From 6136b8a79a7f0ce5e35dc07cd88965953bd0e1af Mon Sep 17 00:00:00 2001
From: Paul Buetow <paul@buetow.org>
Date: Sun, 1 Mar 2026 17:39:53 +0200
Subject: Extract generate::extract_title helper to eliminate duplication

Title extraction ($SED pattern + quote sanitization) was duplicated
in generate, atomfeed, gemfeed, notes, and template modules. Now
centralized with unit tests for heading extraction and empty files.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 lib/atomfeed.source.sh |  2 +-
 lib/gemfeed.source.sh  |  3 +--
 lib/generate.source.sh | 18 +++++++++++++++++-
 lib/notes.source.sh    |  3 +--
 lib/template.source.sh |  2 +-
 5 files changed, 21 insertions(+), 7 deletions(-)

(limited to 'lib')

diff --git a/lib/atomfeed.source.sh b/lib/atomfeed.source.sh
index 3fe46be..20a87d0 100644
--- a/lib/atomfeed.source.sh
+++ b/lib/atomfeed.source.sh
@@ -121,7 +121,7 @@ atomfeed::_entry () {
     assert::not_empty content "$content"
 
     # Extract first heading as post title.
-    local title=$($SED -n '/^# / { s/# //; p; q; }' "$gemfeed_dir/$gmi_file" | tr '"' "'")
+    local title=$(generate::extract_title "$gemfeed_dir/$gmi_file")
     assert::not_empty title "$title"
 
     # Extract first paragraph from Gemtext as the summary.
diff --git a/lib/gemfeed.source.sh b/lib/gemfeed.source.sh
index 40792cc..8700ac4 100644
--- a/lib/gemfeed.source.sh
+++ b/lib/gemfeed.source.sh
@@ -47,8 +47,7 @@ GEMFEED
 
     while read -r gmi_file; do
         # Extract first heading as post title.
-        local title=$($SED -n '/^# / { s/# //; p; q; }' \
-            "$gemfeed_dir/$gmi_file" | tr '"' "'")
+        local title=$(generate::extract_title "$gemfeed_dir/$gmi_file")
 
         # Extract the date from the file name, and also get the word count.
         local filename_date=$(basename "$gemfeed_dir/$gmi_file" | cut -d- -f1,2,3)
diff --git a/lib/generate.source.sh b/lib/generate.source.sh
index 544ae08..17322e7 100644
--- a/lib/generate.source.sh
+++ b/lib/generate.source.sh
@@ -53,6 +53,13 @@ generate::safe_overwrite () {
     fi
 }
 
+# Extract the first heading from a .gmi file as a title, with double quotes
+# replaced by single quotes for safe embedding in feeds and HTML attributes.
+generate::extract_title () {
+    local -r file="$1"; shift
+    $SED -n '/^# / { s/# //; p; q; }' "$file" | tr '"' "'"
+}
+
 # Add other docs (e.g. images, videos) from Gemtext to output format.
 # Skips copying if the output file already exists and is newer than the source.
 generate::fromgmi_add_docs () {
@@ -119,7 +126,7 @@ generate::_to_output_format () {
     dest=${dest/.gmi/.$format}
     local dest_dir=$(dirname "$dest")
 
-    local title=$($SED -n '/^# / { s/# //; p; q; }' "$src" | tr '"' "'")
+    local title=$(generate::extract_title "$src")
     if [[ -z "$title" ]]; then
         title="$SUBTITLE"
     fi
@@ -337,5 +344,14 @@ generate::test () {
     assert::equals "$(cat "$tmp_dir/file")" 'different content'
     assert::equals "$(test -f "$tmp_dir/file.tmp" && echo exists || echo gone)" 'gone'
 
+    # Test generate::extract_title: extracts first heading and sanitizes quotes
+    echo '# My "Great" Title' > "$tmp_dir/test.gmi"
+    echo '## Not this one' >> "$tmp_dir/test.gmi"
+    assert::equals "$(generate::extract_title "$tmp_dir/test.gmi")" "My 'Great' Title"
+
+    # Test generate::extract_title: file with no heading returns empty
+    echo 'Just a paragraph' > "$tmp_dir/noheading.gmi"
+    assert::equals "$(generate::extract_title "$tmp_dir/noheading.gmi")" ''
+
     rm -rf "$tmp_dir"
 }
diff --git a/lib/notes.source.sh b/lib/notes.source.sh
index dc62caf..64625b4 100644
--- a/lib/notes.source.sh
+++ b/lib/notes.source.sh
@@ -27,8 +27,7 @@ NOTES
 
     while read -r gmi_file; do
         # Extract first heading as post title.
-        local title=$($SED -n '/^# / { s/# //; p; q; }' \
-            "$notes_dir/$gmi_file" | tr '"' "'")
+        local title=$(generate::extract_title "$notes_dir/$gmi_file")
         echo "=> ./$gmi_file $title" >> "$notes_dir/index.gmi.tmp"
     done < <(notes::_get_notes)
 
diff --git a/lib/template.source.sh b/lib/template.source.sh
index b93303e..0e34a87 100644
--- a/lib/template.source.sh
+++ b/lib/template.source.sh
@@ -143,7 +143,7 @@ template::inline::_index () {
     for topic in "$@"; do 
         while read -r gmi_file; do
             local date=$(cut -d- -f1,2,3 <<< "$gmi_file")
-            local title=$($SED -n "/^# / { s/# //; p; q; }" "$gmi_file")
+            local title=$(generate::extract_title "$gmi_file")
 
             local current=''
             if [ "$gmi_file" = "$CURRENT_GMI" ]; then
-- 
cgit v1.2.3