mutt_thread.c File Reference

Create/manipulate threading in emails. More...

#include "config.h"
#include <limits.h>
#include <stdbool.h>
#include <stdlib.h>
#include <string.h>
#include <time.h>
#include "mutt/lib.h"
#include "config/lib.h"
#include "email/lib.h"
#include "core/lib.h"
#include "mutt.h"
#include "mutt_thread.h"
#include "context.h"
#include "mutt_menu.h"
#include "mx.h"
#include "protos.h"
#include "sort.h"

Include dependency graph for mutt_thread.c:

Go to the source code of this file.

Macros
#define	CHECK_LIMIT   (!ctx->pattern || e_cur->limited)

Functions
static bool	is_visible (struct Email *e, struct Context *ctx)
	Is the message visible? More...
static bool	need_display_subject (struct Context *ctx, struct Email *e)
	Determines whether to display a message's subject. More...
static void	linearize_tree (struct Context *ctx)
	Flatten an email thread. More...
static void	calculate_visibility (struct Context *ctx, int *max_depth)
	Are tree nodes visible. More...
void	mutt_draw_tree (struct Context *ctx)
	Draw a tree of threaded emails. More...
static void	make_subject_list (struct ListHead *subjects, struct MuttThread *cur, time_t *dateptr)
	Create a sorted list of all subjects in a thread. More...
static struct MuttThread *	find_subject (struct Mailbox *m, struct MuttThread *cur)
	Find the best possible match for a parent based on subject. More...
static struct HashTable *	make_subj_hash (struct Mailbox *m)
	Create a Hash Table for the email subjects. More...
static void	pseudo_threads (struct Context *ctx)
	Thread messages by subject. More...
void	mutt_clear_threads (struct Context *ctx)
	Clear the threading of message in a mailbox. More...
static int	compare_threads (const void *a, const void *b)
	Sorting function for email threads. More...
struct MuttThread *	mutt_sort_subthreads (struct MuttThread *thread, bool init)
	Sort the children of a thread. More...
static void	check_subjects (struct Mailbox *m, bool init)
	Find out which emails' subjects differ from their parent's. More...
void	mutt_sort_threads (struct Context *ctx, bool init)
	Sort email threads. More...
int	mutt_aside_thread (struct Email *e, bool forwards, bool subthreads)
	Find the next/previous (sub)thread. More...
int	mutt_parent_message (struct Context *ctx, struct Email *e, bool find_root)
	Find the parent of a message. More...
void	mutt_set_vnum (struct Context *ctx)
	Set the virtual index number of all the messages in a mailbox. More...
int	mutt_traverse_thread (struct Context *ctx, struct Email *e_cur, MuttThreadFlags flag)
	Recurse through an email thread, matching messages. More...
int	mutt_messages_in_thread (struct Mailbox *m, struct Email *e, int flag)
	Count the messages in a thread. More...
struct HashTable *	mutt_make_id_hash (struct Mailbox *m)
	Create a Hash Table for message-ids. More...
static bool	link_threads (struct Email *parent, struct Email *child, struct Mailbox *m)
	Forcibly link messages together. More...
bool	mutt_link_threads (struct Email *parent, struct EmailList *children, struct Mailbox *m)
	Forcibly link threads together. More...

Variables
bool	C_DuplicateThreads
	Config: Highlight messages with duplicated message IDs. More...
bool	C_HideLimited
	Config: Don't indicate hidden messages, in the thread tree. More...
bool	C_HideMissing
	Config: Don't indicate missing messages, in the thread tree. More...
bool	C_HideThreadSubject
	Config: Hide subjects that are similar to that of the parent message. More...
bool	C_HideTopLimited
	Config: Don't indicate hidden top message, in the thread tree. More...
bool	C_HideTopMissing
	Config: Don't indicate missing top message, in the thread tree. More...
bool	C_NarrowTree
	Config: Draw a narrower thread tree in the index. More...
bool	C_SortRe
	Config: Sort method for the sidebar. More...
bool	C_StrictThreads
	Config: Thread messages using 'In-Reply-To' and 'References' headers. More...
bool	C_ThreadReceived
	Config: Sort threaded messages by their received date. More...

Detailed Description

Create/manipulate threading in emails.

Authors

• Michael R. Elkins

Copyright

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 2 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program. If not, see http://www.gnu.org/licenses/.

Definition in file mutt_thread.c.

Macro Definition Documentation

CHECK_LIMIT

#define CHECK_LIMIT   (!ctx->pattern || e_cur->limited)

Function Documentation

is_visible()

static bool is_visible ( struct Email *  e, struct Context *  ctx  )

static

Is the message visible?

Parameters

e	Email
ctx	Mailbox

Return values

true	If the message is not hidden in some way

Definition at line 65 of file mutt_thread.c.

{
  return e->vnum >= 0 || (e->collapsed && (!ctx->pattern || e->limited));
}

Here is the caller graph for this function:

need_display_subject()

static bool need_display_subject ( struct Context *  ctx, struct Email *  e  )

static

Determines whether to display a message's subject.

Parameters

ctx	Mailbox
e	Email

Return values

true	If the subject should be displayed

Definition at line 76 of file mutt_thread.c.

{
  struct MuttThread *tmp = NULL;
  struct MuttThread *tree = e->thread;

  /* if the user disabled subject hiding, display it */
  if (!C_HideThreadSubject)
    return true;

  /* if our subject is different from our parent's, display it */
  if (e->subject_changed)
    return true;

  /* if our subject is different from that of our closest previously displayed
   * sibling, display the subject */
  for (tmp = tree->prev; tmp; tmp = tmp->prev)
  {
    e = tmp->message;
    if (e && is_visible(e, ctx))
    {
      if (e->subject_changed)
        return true;
      break;
    }
  }

  /* if there is a parent-to-child subject change anywhere between us and our
   * closest displayed ancestor, display the subject */
  for (tmp = tree->parent; tmp; tmp = tmp->parent)
  {
    e = tmp->message;
    if (e)
    {
      if (is_visible(e, ctx))
        return false;
      if (e->subject_changed)
        return true;
    }
  }

  /* if we have no visible parent or previous sibling, display the subject */
  return true;
}

Here is the call graph for this function:

Here is the caller graph for this function:

linearize_tree()

static void linearize_tree ( struct Context *  ctx )

static

Flatten an email thread.

Parameters

ctx

Mailbox

Definition at line 124 of file mutt_thread.c.

{
  if (!ctx || !ctx->mailbox)
    return;

  struct Mailbox *m = ctx->mailbox;

  struct MuttThread *tree = ctx->tree;
  struct Email **array = m->emails + ((C_Sort & SORT_REVERSE) ? m->msg_count - 1 : 0);

  while (tree)
  {
    while (!tree->message)
      tree = tree->child;

    *array = tree->message;
    array += (C_Sort & SORT_REVERSE) ? -1 : 1;

    if (tree->child)
      tree = tree->child;
    else
    {
      while (tree)
      {
        if (tree->next)
        {
          tree = tree->next;
          break;
        }
        else
          tree = tree->parent;
      }
    }
  }
}

Here is the caller graph for this function:

calculate_visibility()

static void calculate_visibility ( struct Context *  ctx, int *  max_depth  )

static

Are tree nodes visible.

Parameters

ctx	Mailbox
max_depth	Maximum depth to check

this calculates whether a node is the root of a subtree that has visible nodes, whether a node itself is visible, whether, if invisible, it has depth anyway, and whether any of its later siblings are roots of visible subtrees. while it's at it, it frees the old thread display, so we can skip parts of the tree in mutt_draw_tree() if we've decided here that we don't care about them any more.

Definition at line 172 of file mutt_thread.c.

{
  struct MuttThread *tmp = NULL;
  struct MuttThread *tree = ctx->tree;
  int hide_top_missing = C_HideTopMissing && !C_HideMissing;
  int hide_top_limited = C_HideTopLimited && !C_HideLimited;
  int depth = 0;

  /* we walk each level backwards to make it easier to compute next_subtree_visible */
  while (tree->next)
    tree = tree->next;
  *max_depth = 0;

  while (true)
  {
    if (depth > *max_depth)
      *max_depth = depth;

    tree->subtree_visible = 0;
    if (tree->message)
    {
      FREE(&tree->message->tree);
      if (is_visible(tree->message, ctx))
      {
        tree->deep = true;
        tree->visible = true;
        tree->message->display_subject = need_display_subject(ctx, tree->message);
        for (tmp = tree; tmp; tmp = tmp->parent)
        {
          if (tmp->subtree_visible)
          {
            tmp->deep = true;
            tmp->subtree_visible = 2;
            break;
          }
          else
            tmp->subtree_visible = 1;
        }
      }
      else
      {
        tree->visible = false;
        tree->deep = !C_HideLimited;
      }
    }
    else
    {
      tree->visible = false;
      tree->deep = !C_HideMissing;
    }
    tree->next_subtree_visible =
        tree->next && (tree->next->next_subtree_visible || tree->next->subtree_visible);
    if (tree->child)
    {
      depth++;
      tree = tree->child;
      while (tree->next)
        tree = tree->next;
    }
    else if (tree->prev)
      tree = tree->prev;
    else
    {
      while (tree && !tree->prev)
      {
        depth--;
        tree = tree->parent;
      }
      if (!tree)
        break;
      tree = tree->prev;
    }
  }

  /* now fix up for the OPTHIDETOP* options if necessary */
  if (hide_top_limited || hide_top_missing)
  {
    tree = ctx->tree;
    while (true)
    {
      if (!tree->visible && tree->deep && (tree->subtree_visible < 2) &&
          ((tree->message && hide_top_limited) || (!tree->message && hide_top_missing)))
      {
        tree->deep = false;
      }
      if (!tree->deep && tree->child && tree->subtree_visible)
        tree = tree->child;
      else if (tree->next)
        tree = tree->next;
      else
      {
        while (tree && !tree->next)
          tree = tree->parent;
        if (!tree)
          break;
        tree = tree->next;
      }
    }
  }
}

Here is the call graph for this function:

Here is the caller graph for this function:

mutt_draw_tree()

void mutt_draw_tree

(

struct Context *

ctx

)

Draw a tree of threaded emails.

Parameters

ctx

Mailbox

Since the graphics characters have a value >255, I have to resort to using escape sequences to pass the information to print_enriched_string(). These are the macros MUTT_TREE_* defined in mutt.h.

ncurses should automatically use the default ASCII characters instead of graphics chars on terminals which don't support them (see the man page for curs_addch).

Definition at line 285 of file mutt_thread.c.

{
  char *pfx = NULL, *mypfx = NULL, *arrow = NULL, *myarrow = NULL, *new_tree = NULL;
  enum TreeChar corner = (C_Sort & SORT_REVERSE) ? MUTT_TREE_ULCORNER : MUTT_TREE_LLCORNER;
  enum TreeChar vtee = (C_Sort & SORT_REVERSE) ? MUTT_TREE_BTEE : MUTT_TREE_TTEE;
  int depth = 0, start_depth = 0, max_depth = 0, width = C_NarrowTree ? 1 : 2;
  struct MuttThread *nextdisp = NULL, *pseudo = NULL, *parent = NULL;
  struct MuttThread *tree = ctx->tree;

  /* Do the visibility calculations and free the old thread chars.
   * From now on we can simply ignore invisible subtrees */
  calculate_visibility(ctx, &max_depth);
  pfx = mutt_mem_malloc((width * max_depth) + 2);
  arrow = mutt_mem_malloc((width * max_depth) + 2);
  while (tree)
  {
    if (depth != 0)
    {
      myarrow = arrow + (depth - start_depth - ((start_depth != 0) ? 0 : 1)) * width;
      if (start_depth == depth)
        myarrow[0] = nextdisp ? MUTT_TREE_LTEE : corner;
      else if (parent->message && !C_HideLimited)
        myarrow[0] = MUTT_TREE_HIDDEN;
      else if (!parent->message && !C_HideMissing)
        myarrow[0] = MUTT_TREE_MISSING;
      else
        myarrow[0] = vtee;
      if (width == 2)
      {
        myarrow[1] = pseudo ? MUTT_TREE_STAR :
                              (tree->duplicate_thread ? MUTT_TREE_EQUALS : MUTT_TREE_HLINE);
      }
      if (tree->visible)
      {
        myarrow[width] = MUTT_TREE_RARROW;
        myarrow[width + 1] = 0;
        new_tree = mutt_mem_malloc(((size_t) depth * width) + 2);
        if (start_depth > 1)
        {
          strncpy(new_tree, pfx, (size_t) width * (start_depth - 1));
          mutt_str_copy(new_tree + (start_depth - 1) * width, arrow,
                        (1 + depth - start_depth) * width + 2);
        }
        else
          mutt_str_copy(new_tree, arrow, ((size_t) depth * width) + 2);
        tree->message->tree = new_tree;
      }
    }
    if (tree->child && (depth != 0))
    {
      mypfx = pfx + (depth - 1) * width;
      mypfx[0] = nextdisp ? MUTT_TREE_VLINE : MUTT_TREE_SPACE;
      if (width == 2)
        mypfx[1] = MUTT_TREE_SPACE;
    }
    parent = tree;
    nextdisp = NULL;
    pseudo = NULL;
    do
    {
      if (tree->child && tree->subtree_visible)
      {
        if (tree->deep)
          depth++;
        if (tree->visible)
          start_depth = depth;
        tree = tree->child;

        /* we do this here because we need to make sure that the first child thread
         * of the old tree that we deal with is actually displayed if any are,
         * or we might set the parent variable wrong while going through it. */
        while (!tree->subtree_visible && tree->next)
          tree = tree->next;
      }
      else
      {
        while (!tree->next && tree->parent)
        {
          if (tree == pseudo)
            pseudo = NULL;
          if (tree == nextdisp)
            nextdisp = NULL;
          if (tree->visible)
            start_depth = depth;
          tree = tree->parent;
          if (tree->deep)
          {
            if (start_depth == depth)
              start_depth--;
            depth--;
          }
        }
        if (tree == pseudo)
          pseudo = NULL;
        if (tree == nextdisp)
          nextdisp = NULL;
        if (tree->visible)
          start_depth = depth;
        tree = tree->next;
        if (!tree)
          break;
      }
      if (!pseudo && tree->fake_thread)
        pseudo = tree;
      if (!nextdisp && tree->next_subtree_visible)
        nextdisp = tree;
    } while (!tree->deep);
  }

  FREE(&pfx);
  FREE(&arrow);
}

Here is the call graph for this function:

Here is the caller graph for this function:

make_subject_list()

static void make_subject_list ( struct ListHead *  subjects, struct MuttThread *  cur, time_t *  dateptr  )

static

Create a sorted list of all subjects in a thread.

Parameters

[out]	subjects	String List of subjects
[in]	cur	Email Thread
[out]	dateptr	Earliest date found in thread

Since we may be trying to attach as a pseudo-thread a MuttThread that has no message, we have to make a list of all the subjects of its most immediate existing descendants.

Definition at line 408 of file mutt_thread.c.

{
  struct MuttThread *start = cur;
  struct Envelope *env = NULL;
  time_t thisdate;
  int rc = 0;

  while (true)
  {
    while (!cur->message)
      cur = cur->child;

    if (dateptr)
    {
      thisdate = C_ThreadReceived ? cur->message->received : cur->message->date_sent;
      if ((*dateptr == 0) || (thisdate < *dateptr))
        *dateptr = thisdate;
    }

    env = cur->message->env;
    if (env->real_subj && ((env->real_subj != env->subject) || (!C_SortRe)))
    {
      struct ListNode *np = NULL;
      STAILQ_FOREACH(np, subjects, entries)
      {
        rc = mutt_str_cmp(env->real_subj, np->data);
        if (rc >= 0)
          break;
      }
      if (!np)
        mutt_list_insert_head(subjects, env->real_subj);
      else if (rc > 0)
        mutt_list_insert_after(subjects, np, env->real_subj);
    }

    while (!cur->next && (cur != start))
    {
      cur = cur->parent;
    }
    if (cur == start)
      break;
    cur = cur->next;
  }
}

Here is the call graph for this function:

Here is the caller graph for this function:

find_subject()

static struct MuttThread* find_subject ( struct Mailbox *  m, struct MuttThread *  cur  )

static

Find the best possible match for a parent based on subject.

Parameters

m	Mailbox
cur	Email to match

Return values

ptr	Best match for a parent

If there are multiple matches, the one which was sent the latest, but before the current message, is used.

Definition at line 462 of file mutt_thread.c.

{
  if (!m)
    return NULL;

  struct HashElem *ptr = NULL;
  struct MuttThread *tmp = NULL, *last = NULL;
  struct ListHead subjects = STAILQ_HEAD_INITIALIZER(subjects);
  time_t date = 0;

  make_subject_list(&subjects, cur, &date);

  struct ListNode *np = NULL;
  STAILQ_FOREACH(np, &subjects, entries)
  {
    for (ptr = mutt_hash_find_bucket(m->subj_hash, np->data); ptr; ptr = ptr->next)
    {
      tmp = ((struct Email *) ptr->data)->thread;
      if ((tmp != cur) &&                  /* don't match the same message */
          !tmp->fake_thread &&             /* don't match pseudo threads */
          tmp->message->subject_changed && /* only match interesting replies */
          !is_descendant(tmp, cur) &&      /* don't match in the same thread */
          (date >= (C_ThreadReceived ? tmp->message->received : tmp->message->date_sent)) &&
          (!last || (C_ThreadReceived ?
                         (last->message->received < tmp->message->received) :
                         (last->message->date_sent < tmp->message->date_sent))) &&
          tmp->message->env->real_subj &&
          mutt_str_equal(np->data, tmp->message->env->real_subj))
      {
        last = tmp; /* best match so far */
      }
    }
  }

  mutt_list_clear(&subjects);
  return last;
}

Here is the call graph for this function:

Here is the caller graph for this function:

make_subj_hash()

static struct HashTable* make_subj_hash ( struct Mailbox *  m )

static

Create a Hash Table for the email subjects.

Parameters

m

Mailbox

Return values

ptr	Newly allocated Hash Table

Definition at line 505 of file mutt_thread.c.

{
  if (!m)
    return NULL;

  struct HashTable *hash = mutt_hash_new(m->msg_count * 2, MUTT_HASH_ALLOW_DUPS);

  for (int i = 0; i < m->msg_count; i++)
  {
    struct Email *e = m->emails[i];
    if (!e || !e->env)
      continue;
    if (e->env->real_subj)
      mutt_hash_insert(hash, e->env->real_subj, e);
  }

  return hash;
}

Here is the call graph for this function:

Here is the caller graph for this function:

pseudo_threads()

static void pseudo_threads ( struct Context *  ctx )

static

Thread messages by subject.

Parameters

ctx

Mailbox

Thread by subject things that didn't get threaded by message-id

Definition at line 530 of file mutt_thread.c.

{
  if (!ctx || !ctx->mailbox)
    return;

  struct Mailbox *m = ctx->mailbox;

  struct MuttThread *tree = ctx->tree;
  struct MuttThread *top = tree;
  struct MuttThread *tmp = NULL, *cur = NULL, *parent = NULL, *curchild = NULL,
                    *nextchild = NULL;

  if (!m->subj_hash)
    m->subj_hash = make_subj_hash(ctx->mailbox);

  while (tree)
  {
    cur = tree;
    tree = tree->next;
    parent = find_subject(ctx->mailbox, cur);
    if (parent)
    {
      cur->fake_thread = true;
      unlink_message(&top, cur);
      insert_message(&parent->child, parent, cur);
      parent->sort_children = true;
      tmp = cur;
      while (true)
      {
        while (!tmp->message)
          tmp = tmp->child;

        /* if the message we're attaching has pseudo-children, they
         * need to be attached to its parent, so move them up a level.
         * but only do this if they have the same real subject as the
         * parent, since otherwise they rightly belong to the message
         * we're attaching. */
        if ((tmp == cur) || mutt_str_equal(tmp->message->env->real_subj,
                                           parent->message->env->real_subj))
        {
          tmp->message->subject_changed = false;

          for (curchild = tmp->child; curchild;)
          {
            nextchild = curchild->next;
            if (curchild->fake_thread)
            {
              unlink_message(&tmp->child, curchild);
              insert_message(&parent->child, parent, curchild);
            }
            curchild = nextchild;
          }
        }

        while (!tmp->next && (tmp != cur))
        {
          tmp = tmp->parent;
        }
        if (tmp == cur)
          break;
        tmp = tmp->next;
      }
    }
  }
  ctx->tree = top;
}

Here is the call graph for this function:

Here is the caller graph for this function:

mutt_clear_threads()

void mutt_clear_threads

(

struct Context *

ctx

)

Clear the threading of message in a mailbox.

Parameters

ctx

Mailbox

Definition at line 601 of file mutt_thread.c.

{
  if (!ctx || !ctx->mailbox || !ctx->mailbox->emails)
    return;

  struct Mailbox *m = ctx->mailbox;

  for (int i = 0; i < m->msg_count; i++)
  {
    struct Email *e = m->emails[i];
    if (!e)
      break;

    /* mailbox may have been only partially read */
    e->thread = NULL;
    e->threaded = false;
  }
  ctx->tree = NULL;

  mutt_hash_free(&ctx->thread_hash);
}

Here is the call graph for this function:

Here is the caller graph for this function:

compare_threads()

static int compare_threads ( const void *  a, const void *  b  )

static

Sorting function for email threads.

Parameters

a	First thread to compare
b	Second thread to compare

Return values

-1	a precedes b
0	a and b are identical
1	b precedes a

Definition at line 631 of file mutt_thread.c.

{
  static sort_t sort_func = NULL;

  if (a && b)
  {
    return (*sort_func)(&(*((struct MuttThread const *const *) a))->sort_key,
                        &(*((struct MuttThread const *const *) b))->sort_key);
  }
  /* a hack to let us reset sort_func even though we can't
   * have extra arguments because of qsort */
  else
  {
    sort_func = mutt_get_sort_func(C_Sort & SORT_MASK);
    return sort_func ? 1 : 0;
  }
}

Here is the call graph for this function:

Here is the caller graph for this function:

mutt_sort_subthreads()

struct MuttThread* mutt_sort_subthreads	(	struct MuttThread *	thread,
		bool	init
	)

Sort the children of a thread.

Parameters

thread	Thread to start at
init	If true, rebuild the thread

Return values

ptr	Sorted threads

Definition at line 655 of file mutt_thread.c.

{
  struct MuttThread **array = NULL, *sort_key = NULL, *top = NULL, *tmp = NULL;
  struct Email *oldsort_key = NULL;
  int i, array_size, sort_top = 0;

  /* we put things into the array backwards to save some cycles,
   * but we want to have to move less stuff around if we're
   * resorting, so we sort backwards and then put them back
   * in reverse order so they're forwards */
  C_Sort ^= SORT_REVERSE;
  if (compare_threads(NULL, NULL) == 0)
    return thread;

  top = thread;

  array_size = 256;
  array = mutt_mem_calloc(array_size, sizeof(struct MuttThread *));
  while (true)
  {
    if (init || !thread->sort_key)
    {
      thread->sort_key = NULL;

      if (thread->parent)
        thread->parent->sort_children = true;
      else
        sort_top = 1;
    }

    if (thread->child)
    {
      thread = thread->child;
      continue;
    }
    else
    {
      /* if it has no children, it must be real. sort it on its own merits */
      thread->sort_key = thread->message;

      if (thread->next)
      {
        thread = thread->next;
        continue;
      }
    }

    while (!thread->next)
    {
      /* if it has siblings and needs to be sorted, sort it... */
      if (thread->prev && (thread->parent ? thread->parent->sort_children : sort_top))
      {
        /* put them into the array */
        for (i = 0; thread; i++, thread = thread->prev)
        {
          if (i >= array_size)
            mutt_mem_realloc(&array, (array_size *= 2) * sizeof(struct MuttThread *));

          array[i] = thread;
        }

        qsort((void *) array, i, sizeof(struct MuttThread *), *compare_threads);

        /* attach them back together.  make thread the last sibling. */
        thread = array[0];
        thread->next = NULL;
        array[i - 1]->prev = NULL;

        if (thread->parent)
          thread->parent->child = array[i - 1];
        else
          top = array[i - 1];

        while (--i)
        {
          array[i - 1]->prev = array[i];
          array[i]->next = array[i - 1];
        }
      }

      if (thread->parent)
      {
        tmp = thread;
        thread = thread->parent;

        if (!thread->sort_key || thread->sort_children)
        {
          /* make sort_key the first or last sibling, as appropriate */
          sort_key = ((!(C_Sort & SORT_LAST)) ^ (!(C_Sort & SORT_REVERSE))) ?
                         thread->child :
                         tmp;

          /* we just sorted its children */
          thread->sort_children = false;

          oldsort_key = thread->sort_key;
          thread->sort_key = thread->message;

          if (C_Sort & SORT_LAST)
          {
            if (!thread->sort_key ||
                ((((C_Sort & SORT_REVERSE) ? 1 : -1) *
                  compare_threads((void *) &thread, (void *) &sort_key)) > 0))
            {
              thread->sort_key = sort_key->sort_key;
            }
          }
          else if (!thread->sort_key)
            thread->sort_key = sort_key->sort_key;

          /* if its sort_key has changed, we need to resort it and siblings */
          if (oldsort_key != thread->sort_key)
          {
            if (thread->parent)
              thread->parent->sort_children = true;
            else
              sort_top = 1;
          }
        }
      }
      else
      {
        C_Sort ^= SORT_REVERSE;
        FREE(&array);
        return top;
      }
    }

    thread = thread->next;
  }
}

Here is the call graph for this function:

Here is the caller graph for this function:

check_subjects()

static void check_subjects ( struct Mailbox *  m, bool  init  )

static

Find out which emails' subjects differ from their parent's.

Parameters

m	Mailbox
init	If true, rebuild the thread

Definition at line 792 of file mutt_thread.c.

{
  if (!m)
    return;

  for (int i = 0; i < m->msg_count; i++)
  {
    struct Email *e = m->emails[i];
    if (!e || !e->thread)
      continue;

    if (e->thread->check_subject)
      e->thread->check_subject = false;
    else if (!init)
      continue;

    /* figure out which messages have subjects different than their parents' */
    struct MuttThread *tmp = e->thread->parent;
    while (tmp && !tmp->message)
    {
      tmp = tmp->parent;
    }

    if (!tmp)
      e->subject_changed = true;
    else if (e->env->real_subj && tmp->message->env->real_subj)
    {
      e->subject_changed = !mutt_str_equal(e->env->real_subj, tmp->message->env->real_subj);
    }
    else
    {
      e->subject_changed = (e->env->real_subj || tmp->message->env->real_subj);
    }
  }
}

Here is the call graph for this function:

Here is the caller graph for this function:

mutt_sort_threads()

void mutt_sort_threads	(	struct Context *	ctx,
		bool	init
	)

Sort email threads.

Parameters

ctx	Mailbox
init	If true, rebuild the thread

Definition at line 833 of file mutt_thread.c.

{
  if (!ctx || !ctx->mailbox)
    return;

  struct Mailbox *m = ctx->mailbox;

  struct Email *e = NULL;
  int i, oldsort, using_refs = 0;
  struct MuttThread *thread = NULL, *tnew = NULL, *tmp = NULL;
  struct MuttThread top = { 0 };
  struct ListNode *ref = NULL;

  /* Set C_Sort to the secondary method to support the set sort_aux=reverse-*
   * settings.  The sorting functions just look at the value of SORT_REVERSE */
  oldsort = C_Sort;
  C_Sort = C_SortAux;

  if (!ctx->thread_hash)
    init = true;

  if (init)
  {
    ctx->thread_hash = mutt_hash_new(m->msg_count * 2, MUTT_HASH_ALLOW_DUPS);
    mutt_hash_set_destructor(ctx->thread_hash, thread_hash_destructor, 0);
  }

  /* we want a quick way to see if things are actually attached to the top of the
   * thread tree or if they're just dangling, so we attach everything to a top
   * node temporarily */
  top.parent = NULL;
  top.next = NULL;
  top.prev = NULL;

  top.child = ctx->tree;
  for (thread = ctx->tree; thread; thread = thread->next)
    thread->parent = ⊤

  /* put each new message together with the matching messageless MuttThread if it
   * exists.  otherwise, if there is a MuttThread that already has a message, thread
   * new message as an identical child.  if we didn't attach the message to a
   * MuttThread, make a new one for it. */
  for (i = 0; i < m->msg_count; i++)
  {
    e = m->emails[i];
    if (!e)
      continue;

    if (!e->thread)
    {
      if ((!init || C_DuplicateThreads) && e->env->message_id)
        thread = mutt_hash_find(ctx->thread_hash, e->env->message_id);
      else
        thread = NULL;

      if (thread && !thread->message)
      {
        /* this is a message which was missing before */
        thread->message = e;
        e->thread = thread;
        thread->check_subject = true;

        /* mark descendants as needing subject_changed checked */
        for (tmp = (thread->child ? thread->child : thread); tmp != thread;)
        {
          while (!tmp->message)
            tmp = tmp->child;
          tmp->check_subject = true;
          while (!tmp->next && (tmp != thread))
            tmp = tmp->parent;
          if (tmp != thread)
            tmp = tmp->next;
        }

        if (thread->parent)
        {
          /* remove threading info above it based on its children, which we'll
           * recalculate based on its headers.  make sure not to leave
           * dangling missing messages.  note that we haven't kept track
           * of what info came from its children and what from its siblings'
           * children, so we just remove the stuff that's definitely from it */
          do
          {
            tmp = thread->parent;
            unlink_message(&tmp->child, thread);
            thread->parent = NULL;
            thread->sort_key = NULL;
            thread->fake_thread = false;
            thread = tmp;
          } while (thread != &top && !thread->child && !thread->message);
        }
      }
      else
      {
        tnew = (C_DuplicateThreads ? thread : NULL);

        thread = mutt_mem_calloc(1, sizeof(struct MuttThread));
        thread->message = e;
        thread->check_subject = true;
        e->thread = thread;
        mutt_hash_insert(ctx->thread_hash,
                         e->env->message_id ? e->env->message_id : "", thread);

        if (tnew)
        {
          if (tnew->duplicate_thread)
            tnew = tnew->parent;

          thread = e->thread;

          insert_message(&tnew->child, tnew, thread);
          thread->duplicate_thread = true;
          thread->message->threaded = true;
        }
      }
    }
    else
    {
      /* unlink pseudo-threads because they might be children of newly
       * arrived messages */
      thread = e->thread;
      for (tnew = thread->child; tnew;)
      {
        tmp = tnew->next;
        if (tnew->fake_thread)
        {
          unlink_message(&thread->child, tnew);
          insert_message(&top.child, &top, tnew);
          tnew->fake_thread = false;
        }
        tnew = tmp;
      }
    }
  }

  /* thread by references */
  for (i = 0; i < m->msg_count; i++)
  {
    e = m->emails[i];
    if (!e)
      break;

    if (e->threaded)
      continue;
    e->threaded = true;

    thread = e->thread;
    if (!thread)
      continue;
    using_refs = 0;

    while (true)
    {
      if (using_refs == 0)
      {
        /* look at the beginning of in-reply-to: */
        ref = STAILQ_FIRST(&e->env->in_reply_to);
        if (ref)
          using_refs = 1;
        else
        {
          ref = STAILQ_FIRST(&e->env->references);
          using_refs = 2;
        }
      }
      else if (using_refs == 1)
      {
        /* if there's no references header, use all the in-reply-to:
         * data that we have.  otherwise, use the first reference
         * if it's different than the first in-reply-to, otherwise use
         * the second reference (since at least eudora puts the most
         * recent reference in in-reply-to and the rest in references) */
        if (STAILQ_EMPTY(&e->env->references))
          ref = STAILQ_NEXT(ref, entries);
        else
        {
          if (!mutt_str_equal(ref->data, STAILQ_FIRST(&e->env->references)->data))
            ref = STAILQ_FIRST(&e->env->references);
          else
            ref = STAILQ_NEXT(STAILQ_FIRST(&e->env->references), entries);

          using_refs = 2;
        }
      }
      else
        ref = STAILQ_NEXT(ref, entries); /* go on with references */

      if (!ref)
        break;

      tnew = mutt_hash_find(ctx->thread_hash, ref->data);
      if (tnew)
      {
        if (tnew->duplicate_thread)
          tnew = tnew->parent;
        if (is_descendant(tnew, thread)) /* no loops! */
          continue;
      }
      else
      {
        tnew = mutt_mem_calloc(1, sizeof(struct MuttThread));
        mutt_hash_insert(ctx->thread_hash, ref->data, tnew);
      }

      if (thread->parent)
        unlink_message(&top.child, thread);
      insert_message(&tnew->child, tnew, thread);
      thread = tnew;
      if (thread->message || (thread->parent && (thread->parent != &top)))
        break;
    }

    if (!thread->parent)
      insert_message(&top.child, &top, thread);
  }

  /* detach everything from the temporary top node */
  for (thread = top.child; thread; thread = thread->next)
  {
    thread->parent = NULL;
  }
  ctx->tree = top.child;

  check_subjects(ctx->mailbox, init);

  if (!C_StrictThreads)
    pseudo_threads(ctx);

  if (ctx->tree)
  {
    ctx->tree = mutt_sort_subthreads(ctx->tree, init);

    /* restore the oldsort order. */
    C_Sort = oldsort;

    /* Put the list into an array. */
    linearize_tree(ctx);

    /* Draw the thread tree. */
    mutt_draw_tree(ctx);
  }
}

Here is the call graph for this function:

Here is the caller graph for this function:

mutt_aside_thread()

int mutt_aside_thread	(	struct Email *	e,
		bool	forwards,
		bool	subthreads
	)

Find the next/previous (sub)thread.

Parameters

e	Search from this Email
forwards	Direction to search: 'true' forwards, 'false' backwards
subthreads	Search subthreads: 'true' subthread, 'false' not

Return values

num	Index into the virtual email table

Definition at line 1083 of file mutt_thread.c.

{
  struct MuttThread *cur = NULL;
  struct Email *e_tmp = NULL;

  if ((C_Sort & SORT_MASK) != SORT_THREADS)
  {
    mutt_error(_("Threading is not enabled"));
    return e->vnum;
  }

  cur = e->thread;

  if (subthreads)
  {
    if (forwards ^ ((C_Sort & SORT_REVERSE) != 0))
    {
      while (!cur->next && cur->parent)
        cur = cur->parent;
    }
    else
    {
      while (!cur->prev && cur->parent)
        cur = cur->parent;
    }
  }
  else
  {
    while (cur->parent)
      cur = cur->parent;
  }

  if (forwards ^ ((C_Sort & SORT_REVERSE) != 0))
  {
    do
    {
      cur = cur->next;
      if (!cur)
        return -1;
      e_tmp = find_virtual(cur, 0);
    } while (!e_tmp);
  }
  else
  {
    do
    {
      cur = cur->prev;
      if (!cur)
        return -1;
      e_tmp = find_virtual(cur, 1);
    } while (!e_tmp);
  }

  return e_tmp->vnum;
}

Here is the call graph for this function:

mutt_parent_message()

int mutt_parent_message	(	struct Context *	ctx,
		struct Email *	e,
		bool	find_root
	)

Find the parent of a message.

Parameters

ctx	Mailbox
e	Current Email
find_root	If true, find the root message

Return values

>=0	Virtual index number of parent/root message
-1	Error

Definition at line 1147 of file mutt_thread.c.

{
  if (!ctx || !e)
    return -1;

  struct MuttThread *thread = NULL;
  struct Email *e_parent = NULL;

  if ((C_Sort & SORT_MASK) != SORT_THREADS)
  {
    mutt_error(_("Threading is not enabled"));
    return e->vnum;
  }

  /* Root may be the current message */
  if (find_root)
    e_parent = e;

  for (thread = e->thread->parent; thread; thread = thread->parent)
  {
    e = thread->message;
    if (e)
    {
      e_parent = e;
      if (!find_root)
        break;
    }
  }

  if (!e_parent)
  {
    mutt_error(_("Parent message is not available"));
    return -1;
  }
  if (!is_visible(e_parent, ctx))
  {
    if (find_root)
      mutt_error(_("Root message is not visible in this limited view"));
    else
      mutt_error(_("Parent message is not visible in this limited view"));
    return -1;
  }
  return e_parent->vnum;
}

Here is the call graph for this function:

Here is the caller graph for this function:

mutt_set_vnum()

void mutt_set_vnum

(

struct Context *

ctx

)

Set the virtual index number of all the messages in a mailbox.

Parameters

ctx

Mailbox

Definition at line 1196 of file mutt_thread.c.

{
  if (!ctx || !ctx->mailbox)
    return;

  struct Mailbox *m = ctx->mailbox;

  struct Email *e = NULL;

  m->vcount = 0;
  ctx->vsize = 0;
  int padding = mx_msg_padding_size(m);

  for (int i = 0; i < m->msg_count; i++)
  {
    e = m->emails[i];
    if (!e)
      break;

    if (e->vnum >= 0)
    {
      e->vnum = m->vcount;
      m->v2r[m->vcount] = i;
      m->vcount++;
      ctx->vsize += e->content->length + e->content->offset - e->content->hdr_offset + padding;
      e->num_hidden = mutt_get_hidden(ctx, e);
    }
  }
}

Here is the call graph for this function:

Here is the caller graph for this function:

mutt_traverse_thread()

int mutt_traverse_thread	(	struct Context *	ctx,
		struct Email *	e_cur,
		MuttThreadFlags	flag
	)

Recurse through an email thread, matching messages.

Parameters

ctx	Mailbox
e_cur	Current Email
flag	Flag to set, see MuttThreadFlags

Return values

num	Number of matches

Definition at line 1233 of file mutt_thread.c.

{
  struct MuttThread *thread = NULL, *top = NULL;
  struct Email *e_root = NULL;
  int final, reverse = (C_Sort & SORT_REVERSE), minmsgno;
  int num_hidden = 0, new_mail = 0, old_mail = 0;
  bool flagged = false;
  int min_unread_msgno = INT_MAX, min_unread = e_cur->vnum;
#define CHECK_LIMIT (!ctx->pattern || e_cur->limited)

  if (((C_Sort & SORT_MASK) != SORT_THREADS) && !(flag & MUTT_THREAD_GET_HIDDEN))
  {
    mutt_error(_("Threading is not enabled"));
    return e_cur->vnum;
  }

  if (!e_cur->thread)
  {
    return e_cur->vnum;
  }

  final = e_cur->vnum;
  thread = e_cur->thread;
  while (thread->parent)
    thread = thread->parent;
  top = thread;
  while (!thread->message)
    thread = thread->child;
  e_cur = thread->message;
  minmsgno = e_cur->msgno;

  if (!e_cur->read && CHECK_LIMIT)
  {
    if (e_cur->old)
      old_mail = 2;
    else
      new_mail = 1;
    if (e_cur->msgno < min_unread_msgno)
    {
      min_unread = e_cur->vnum;
      min_unread_msgno = e_cur->msgno;
    }
  }

  if (e_cur->flagged && CHECK_LIMIT)
    flagged = true;

  if ((e_cur->vnum == -1) && CHECK_LIMIT)
    num_hidden++;

  if (flag & (MUTT_THREAD_COLLAPSE | MUTT_THREAD_UNCOLLAPSE))
  {
    e_cur->pair = 0; /* force index entry's color to be re-evaluated */
    e_cur->collapsed = flag & MUTT_THREAD_COLLAPSE;
    if (e_cur->vnum != -1)
    {
      e_root = e_cur;
      if (flag & MUTT_THREAD_COLLAPSE)
        final = e_root->vnum;
    }
  }

  if ((thread == top) && !(thread = thread->child))
  {
    /* return value depends on action requested */
    if (flag & (MUTT_THREAD_COLLAPSE | MUTT_THREAD_UNCOLLAPSE))
      return final;
    if (flag & MUTT_THREAD_UNREAD)
      return (old_mail && new_mail) ? new_mail : (old_mail ? old_mail : new_mail);
    if (flag & MUTT_THREAD_GET_HIDDEN)
      return num_hidden;
    if (flag & MUTT_THREAD_NEXT_UNREAD)
      return min_unread;
    if (flag & MUTT_THREAD_FLAGGED)
      return flagged;
  }

  while (true)
  {
    e_cur = thread->message;

    if (e_cur)
    {
      if (flag & (MUTT_THREAD_COLLAPSE | MUTT_THREAD_UNCOLLAPSE))
      {
        e_cur->pair = 0; /* force index entry's color to be re-evaluated */
        e_cur->collapsed = flag & MUTT_THREAD_COLLAPSE;
        if (!e_root && CHECK_LIMIT)
        {
          e_root = e_cur;
          if (flag & MUTT_THREAD_COLLAPSE)
            final = e_root->vnum;
        }

        if (reverse && (flag & MUTT_THREAD_COLLAPSE) && (e_cur->msgno < minmsgno) && CHECK_LIMIT)
        {
          minmsgno = e_cur->msgno;
          final = e_cur->vnum;
        }

        if (flag & MUTT_THREAD_COLLAPSE)
        {
          if (e_cur != e_root)
            e_cur->vnum = -1;
        }
        else
        {
          if (CHECK_LIMIT)
            e_cur->vnum = e_cur->msgno;
        }
      }

      if (!e_cur->read && CHECK_LIMIT)
      {
        if (e_cur->old)
          old_mail = 2;
        else
          new_mail = 1;
        if (e_cur->msgno < min_unread_msgno)
        {
          min_unread = e_cur->vnum;
          min_unread_msgno = e_cur->msgno;
        }
      }

      if (e_cur->flagged && CHECK_LIMIT)
        flagged = true;

      if ((e_cur->vnum == -1) && CHECK_LIMIT)
        num_hidden++;
    }

    if (thread->child)
      thread = thread->child;
    else if (thread->next)
      thread = thread->next;
    else
    {
      bool done = false;
      while (!thread->next)
      {
        thread = thread->parent;
        if (thread == top)
        {
          done = true;
          break;
        }
      }
      if (done)
        break;
      thread = thread->next;
    }
  }

  /* return value depends on action requested */
  if (flag & (MUTT_THREAD_COLLAPSE | MUTT_THREAD_UNCOLLAPSE))
    return final;
  if (flag & MUTT_THREAD_UNREAD)
    return (old_mail && new_mail) ? new_mail : (old_mail ? old_mail : new_mail);
  if (flag & MUTT_THREAD_GET_HIDDEN)
    return num_hidden + 1;
  if (flag & MUTT_THREAD_NEXT_UNREAD)
    return min_unread;
  if (flag & MUTT_THREAD_FLAGGED)
    return flagged;

  return 0;
#undef CHECK_LIMIT
}

mutt_messages_in_thread()

int mutt_messages_in_thread	(	struct Mailbox *	m,
		struct Email *	e,
		int	flag
	)

Count the messages in a thread.

Parameters

m	Mailbox
e	Email
flag	Flag, see notes below

Return values

num	Number of message / Our position

If flag is 0, we want to know how many messages are in the thread. If flag is 1, we want to know our position in the thread.

Definition at line 1413 of file mutt_thread.c.

{
  if (!m || !e)
    return 1;

  struct MuttThread *threads[2];
  int rc;

  if (((C_Sort & SORT_MASK) != SORT_THREADS) || !e->thread)
    return 1;

  threads[0] = e->thread;
  while (threads[0]->parent)
    threads[0] = threads[0]->parent;

  threads[1] = flag ? e->thread : threads[0]->next;

  for (int i = 0; i < ((flag || !threads[1]) ? 1 : 2); i++)
  {
    while (!threads[i]->message)
      threads[i] = threads[i]->child;
  }

  if (C_Sort & SORT_REVERSE)
    rc = threads[0]->message->msgno - (threads[1] ? threads[1]->message->msgno : -1);
  else
  {
    rc = (threads[1] ? threads[1]->message->msgno : m->msg_count) -
         threads[0]->message->msgno;
  }

  if (flag)
    rc += 1;

  return rc;
}

Here is the caller graph for this function:

mutt_make_id_hash()

struct HashTable* mutt_make_id_hash

(

struct Mailbox *

m

)

Create a Hash Table for message-ids.

Parameters

m

Mailbox

Return values

ptr	Newly allocated Hash Table

Definition at line 1455 of file mutt_thread.c.

{
  struct HashTable *hash = mutt_hash_new(m->msg_count * 2, MUTT_HASH_NO_FLAGS);

  for (int i = 0; i < m->msg_count; i++)
  {
    struct Email *e = m->emails[i];
    if (!e || !e->env)
      continue;

    if (e->env->message_id)
      mutt_hash_insert(hash, e->env->message_id, e);
  }

  return hash;
}

Here is the call graph for this function:

Here is the caller graph for this function:

link_threads()

static bool link_threads ( struct Email *  parent, struct Email *  child, struct Mailbox *  m  )

static

Forcibly link messages together.

Parameters

parent	Parent Email
child	Child Email
m	Mailbox

Return values

true	On success

Definition at line 1479 of file mutt_thread.c.

{
  if (child == parent)
    return false;

  mutt_break_thread(child);
  mutt_list_insert_head(&child->env->in_reply_to, mutt_str_dup(parent->env->message_id));
  mutt_set_flag(m, child, MUTT_TAG, false);

  child->changed = true;
  child->env->changed |= MUTT_ENV_CHANGED_IRT;
  return true;
}

Here is the call graph for this function:

Here is the caller graph for this function:

mutt_link_threads()

bool mutt_link_threads	(	struct Email *	parent,
		struct EmailList *	children,
		struct Mailbox *	m
	)

Forcibly link threads together.

Parameters

parent	Parent Email
children	List of children Emails
m	Mailbox

Return values

true	On success

Definition at line 1500 of file mutt_thread.c.

{
  if (!parent || !children || !m)
    return false;

  bool changed = false;

  struct EmailNode *en = NULL;
  STAILQ_FOREACH(en, children, entries)
  {
    changed |= link_threads(parent, en->email, m);
  }

  return changed;
}

Here is the call graph for this function:

Here is the caller graph for this function:

Variable Documentation

C_DuplicateThreads

bool C_DuplicateThreads

Config: Highlight messages with duplicated message IDs.

Definition at line 48 of file mutt_thread.c.

C_HideLimited

bool C_HideLimited

Config: Don't indicate hidden messages, in the thread tree.

Definition at line 49 of file mutt_thread.c.

C_HideMissing

bool C_HideMissing

Config: Don't indicate missing messages, in the thread tree.

Definition at line 50 of file mutt_thread.c.

C_HideThreadSubject

bool C_HideThreadSubject

Config: Hide subjects that are similar to that of the parent message.

Definition at line 51 of file mutt_thread.c.

C_HideTopLimited

bool C_HideTopLimited

Config: Don't indicate hidden top message, in the thread tree.

Definition at line 52 of file mutt_thread.c.

C_HideTopMissing

bool C_HideTopMissing

Config: Don't indicate missing top message, in the thread tree.

Definition at line 53 of file mutt_thread.c.

C_NarrowTree

bool C_NarrowTree

Config: Draw a narrower thread tree in the index.

Definition at line 54 of file mutt_thread.c.

C_SortRe

bool C_SortRe

Config: Sort method for the sidebar.

Definition at line 55 of file mutt_thread.c.

C_StrictThreads

bool C_StrictThreads

Config: Thread messages using 'In-Reply-To' and 'References' headers.

Definition at line 56 of file mutt_thread.c.

C_ThreadReceived

bool C_ThreadReceived

Config: Sort threaded messages by their received date.

Definition at line 57 of file mutt_thread.c.