Update backlinks when a page is renamed

otterwiki/backlinks.py

@@ -0,0 +1,258 @@
+#!/usr/bin/env python3
+
+"""
+When a page is renamed this module rewrites all links
+in the wiki that point to the old page so that they point
+to the new page name instead.
+"""
+
+import re
+
+from otterwiki.server import app, storage
+from urllib.parse import quote, unquote
+
+"""Rewrite WikiLinks and images pointing to a renamed page."""
+
+# Matches a single WikiLink and captures its inner content.
+# WikiLink inner content never contains a closing bracket.
+WIKILINK_RE = re.compile(r"\[\[([^\]]+)\]\]")
+
+# Matches nested images inside links
+IMAGE_LINK_WRAPPER_RE = re.compile(r"\[(!\[[^\]]*\]\([^)]+\))\]\((/[^)]+)\)")
+
+# Matches a Markdown image link
+IMAGE_LINK_RE = re.compile(r"!\[([^\]]*)\]\((/[^)]+)\)")
+
+# Matches Markdown links
+# Example: [text](/Parent/ChildPage)
+MARKDOWN_LINK_RE = re.compile(r"\[([^\]]+)\]\((/[^)]+)\)")
+
+
+def _is_linktitle_style():
+    """True if the link is the *left* side of ``[[left|right]]``."""
+    style = (
+        app.config.get("WIKILINK_STYLE", "").upper().replace("_", "").strip()
+    )
+    return style in ("LINKTITLE", "PAGENAMETITLE")
+
+
+def _split_link_title(inner, linktitle_style):
+    """Split a WikiLink's inner content into (link_part, title_part).
+
+    ``title_part`` is None when the link carries no explicit title, in
+    which case the link doubles as the displayed text.
+    """
+    if "|" in inner:
+        left, right = inner.split("|", 1)
+        if linktitle_style:
+            return left, right
+        return right, left
+    # no pipe: the whole content is both link and displayed text
+    return inner, None
+
+
+def _rewrite_inner(inner, old_key, new_pagepath, linktitle_style):
+    """Return the rewritten inner content, or None if it doesn't match."""
+    from otterwiki.helper import get_filename
+
+    link_part, title_part = _split_link_title(inner, linktitle_style)
+
+    # preserve a leading slash (absolute links) and any #anchor
+    leading_slash = link_part.startswith("/")
+    page, hsep, anchor = link_part.strip().lstrip("/").partition("#")
+    page = unquote(page.strip())
+    if not page:
+        # pure anchor link like [[#section]] - nothing to do
+        return None
+
+    # resolve the link target to its on-disk filename and compare.
+    if get_filename(page) != old_key:
+        return None
+
+    new_link = ("/" if leading_slash else "") + new_pagepath
+    if hsep:
+        new_link += "#" + anchor
+
+    if title_part is None:
+        return new_link
+    if linktitle_style:
+        return new_link + "|" + title_part
+    return title_part + "|" + new_link
+
+
+def _rewrite_attachment(match, old_pagepath, new_pagepath):
+    """Rewrite Markdown image links pointing to attachments on a renamed page."""
+
+    alt = match.group(1)
+    url = match.group(2)
+
+    new_url = _rewrite_url(
+        url,
+        old_pagepath,
+        new_pagepath,
+    )
+
+    return f"![{alt}]({new_url})"
+
+
+def _rewrite_markdown_link(match, new_pagepath, old_pagepath):
+    """Rewrite Markdown links pointing to a renamed page."""
+
+    text = match.group(1)
+    url = match.group(2)
+
+    new_url = _rewrite_url(
+        url,
+        old_pagepath,
+        new_pagepath,
+    )
+
+    return f"[{text}]({new_url})"
+
+
+def _rewrite_url(url, old_pagepath, new_pagepath):
+    """Rewrite a Markdown URL pointing to a renamed page or one of its
+    attachments.
+
+    Returns the rewritten URL, or the original URL if no rewrite is needed.
+    """
+
+    from otterwiki.helper import get_filename
+
+    retain_case = app.config.get("RETAIN_PAGE_NAME_CASE", False)
+
+    # Split query string.
+    path, qsep, query = url.partition("?")
+
+    # Split anchor.
+    path, hsep, anchor = path.partition("#")
+
+    #
+    # First: attachment URLs
+    #
+    old_page_dir = quote(old_pagepath.rsplit(".", 1)[0].strip("/"), safe="/")
+    new_page_dir = quote(new_pagepath.rsplit(".", 1)[0].strip("/"), safe="/")
+
+    old_prefix = "/" + old_page_dir + "/"
+
+    if retain_case:
+        matches = path.startswith(old_prefix)
+    else:
+        matches = path.lower().startswith(old_prefix.lower())
+
+    if matches:
+        relative = path[len(old_prefix) - 1 :]
+        new_path = "/" + new_page_dir + relative
+
+    else:
+        #
+        # Page URL
+        #
+        leading_slash = path.startswith("/")
+
+        page = unquote(path.lstrip("/"))
+
+        if not page:
+            return url
+
+        old_key = get_filename(old_pagepath)
+        if get_filename(page) != old_key:
+            return url
+
+        encoded = quote(new_pagepath, safe="/")
+        new_path = ("/" if leading_slash else "") + encoded
+
+    # Restore anchor.
+    if hsep:
+        new_path += "#" + anchor
+
+    # Restore query string.
+    if qsep:
+        new_path += "?" + query
+
+    return new_path
+
+
+def _rewrite_content(
+    content, old_key, new_pagepath, linktitle_style, old_pagepath
+):
+    def repl(match):
+        new_inner = _rewrite_inner(
+            match.group(1), old_key, new_pagepath, linktitle_style
+        )
+        if new_inner is None:
+            return match.group(0)
+        return "[[" + new_inner + "]]"
+
+    def wrapper_repl(match):
+        image = match.group(1)
+        url = _rewrite_url(match.group(2), old_pagepath, new_pagepath)
+        return f"[{image}]({url})"
+
+    def attachment_repl(match):
+        return _rewrite_attachment(match, old_pagepath, new_pagepath)
+
+    def markdown_link_repl(match):
+        return _rewrite_markdown_link(
+            match,
+            new_pagepath,
+            old_pagepath,
+        )
+
+    # Rewrite WikiLinks
+    content = WIKILINK_RE.sub(repl, content)
+
+    # Rewrite nested images inside links
+    content = IMAGE_LINK_WRAPPER_RE.sub(wrapper_repl, content)
+
+    # Rewrite image links
+    content = IMAGE_LINK_RE.sub(attachment_repl, content)
+
+    # Rewrite Markdown links
+    content = MARKDOWN_LINK_RE.sub(markdown_link_repl, content)
+
+    return content
+
+
+def rename_backlinks(old_pagepath, new_pagepath):
+    from otterwiki.helper import get_filename
+
+    old_key = get_filename(old_pagepath)
+    linktitle_style = _is_linktitle_style()
+
+    all_files, _ = storage.list()
+    md_files = [f for f in all_files if f.endswith(".md")]
+
+    updated = {}
+    for md_file in md_files:
+        try:
+            content = storage.load(md_file, mode="r")
+            new_content = _rewrite_content(
+                content, old_key, new_pagepath, linktitle_style, old_pagepath
+            )
+            if new_content == content:
+                continue
+            # storage.update() writes the updated content to file.
+            storage.update(
+                filename=md_file,
+                content=new_content,
+            )
+            updated[md_file] = new_content
+        except Exception as e:
+            app.logger.warning(
+                "Skipping page '%s': %s",
+                md_file,
+                e,
+            )
+            continue
+
+    if updated:
+        app.logger.info(
+            "rename_backlinks: rewrote backlinks in %d page(s) "
+            "after renaming '%s' to '%s'",
+            len(updated),
+            old_pagepath,
+            new_pagepath,
+        )
+
+    return updated

otterwiki/gitstorage.py

@@ -322,16 +322,12 @@ def log_slow(self, filename=None):
         # build and return logfile
         return [self._get_metadata_of_commit(commit) for commit in commits]
 
-    def store(
+    def update(
         self,
         filename,
         content,
-        message: str | None = "",
-        author=("", ""),
         mode="w",
     ):
-        if message is None:
-            message = ""
         dirname = os.path.dirname(filename)
         if dirname != "":
             os.makedirs(
@@ -344,6 +340,22 @@ def store(
         diff = self.repo.index.diff(None, paths=filename)
         if len(diff) == 0 and filename not in self.repo.untracked_files:
             return False
+
+        return True
+
+    def store(
+        self,
+        filename,
+        content,
+        message: str | None = "",
+        author=("", ""),
+        mode="w",
+    ):
+        if not self.update(filename, content, mode):
+            return False
+
+        if message is None:
+            message = ""
         # add and commit to git
         index = self.repo.index
         index.add([filename])
@@ -450,6 +462,7 @@ def rename(
         message=None,
         author=None,
         no_commit=False,
+        files_updated=[],
     ):
         if self.exists(new_filename):
             raise StorageError(
@@ -472,7 +485,8 @@ def rename(
         if message is None:
             message = "{} renamed to {}.".format(old_filename, new_filename)
         if not no_commit:
-            self.commit([new_filename], message, author, no_add=True)
+            files_updated.append(new_filename)
+            self.commit(files_updated, message, author, no_add=True)
 
     def list(self, p=None, depth=None, exclude=[]):
         excludes = [".git"] + exclude

otterwiki/help_admin.md

@@ -95,7 +95,7 @@ setting. Setting this to `optional` will allow empty commit messages.
 
 #### Page case name
 An Otter Wiki stores pages in files with names of all lowercase names. To retain
-the upper and lower case of the filenames, check <span class="help-button"><input type="checkbox" style="display:inline;" id="true-retain-page-name" checked> Retain page name case</span>.
+the upper and lower case of the filenames, check <span class="help-button"><input type="checkbox" style="display:inline;" id="true-retain-page-name" checked>Retain page name case</span>.
 
 #### Git Web server
 With <span class="help-button"><input type="checkbox" style="display:inline;" id="true-git-webserver" checked> Enable Git Server</span> allow users with the permission to READ to clone and pull the wiki content via git and users with UPLOAD/Attachment management permissions to push content. HTTP Basic authentication is used for non anonymous access. There is no option for using git via ssh. When enabled, users find the URL to clone the repositroy in their settings.

otterwiki/static/js/otterwiki.js

@@ -86,6 +86,21 @@ var otterwiki = {
         for (let i = 0; i < checkboxes.length; i++) {
             checkboxes[i].checked = !checkboxes[i].checked;
         }
+    },
+    retain_update_backlinks_checkbox: function() {
+        const checkbox = document.getElementById("update_backlinks");
+        if (!checkbox) return;
+
+        const key = "otterwiki/rename/update-backlinks";
+
+        const saved = sessionStorage.getItem(key);
+        if (saved !== null) {
+            checkbox.checked = (saved === "true");
+        }
+
+        checkbox.addEventListener("change", () => {
+            sessionStorage.setItem(key, checkbox.checked);
+        });
     }
 }
 
@@ -269,6 +284,8 @@ window.addEventListener("keydown", function(event) {
 
 
 document.addEventListener("DOMContentLoaded", () => {
+  otterwiki.retain_update_backlinks_checkbox();
+
   if (document.body.dataset.pageIndexRetainUserExpandedNodes === 'False') return;
 
   const detailsList = document.querySelectorAll("details");

otterwiki/templates/rename.html

@@ -65,6 +65,20 @@
     <label for="new_pagename" class="required w-150">New name</label>
     <input type="text" class="form-control" name="new_pagename" id="new_pagename" value="{{new_pagename if new_pagename else pagename}}" placeholder="{{pagename}}" autofocus="" onfocus="this.setSelectionRange(this.value.length,this.value.length);">
   </div>
+  <div class="form-group">
+    <label for="update_backlinks" class="w-150">Update backlinks</label>
+    <div class="form-check">
+      <input
+        type="checkbox"
+        class="form-check-input"
+        name="update_backlinks"
+        id="update_backlinks"
+        value="1">
+      <label class="form-check-label" for="update_backlinks">
+        Update links pointing to this page
+      </label>
+    </div>
+  </div>
   <div class="form-group">
     <label for="message" class="w-150">Commit message</label>
     <input type="text" class="form-control" name="message" id="message" placeholder="Renamed {{pagename}} to {{new_pagename if new_pagename else "..."}}" value="{{ message if message }}">

otterwiki/views.py

@@ -534,6 +534,7 @@ def rename(path):
             new_pagename=request.form.get("new_pagename"),
             message=request.form.get("message"),
             author=otterwiki.auth.get_author(),
+            update_backlinks=request.form.get("update_backlinks"),
         )
     return p.rename_form()