From 4019ad514b68e44951b0a1d31316e26877467175 Mon Sep 17 00:00:00 2001
From: jaseg <git-bigdata-wsl-arch@jaseg.de>
Date: Mon, 12 Jul 2021 14:01:13 +0200
Subject: Add diff pdf generator

---
 paper/diffinator.py | 78 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 78 insertions(+)
 create mode 100644 paper/diffinator.py

(limited to 'paper/diffinator.py')

diff --git a/paper/diffinator.py b/paper/diffinator.py
new file mode 100644
index 0000000..2972d47
--- /dev/null
+++ b/paper/diffinator.py
@@ -0,0 +1,78 @@
+#!/usr/bin/env python3
+
+import re
+import subprocess
+
+import click
+
+
+@click.command()
+@click.argument('texfile')
+@click.argument('bibliography')
+@click.argument('revision')
+def generate_git_tex_diff(texfile, bibliography, revision):
+    with open(texfile) as f:
+        tex_lines = len(list(f))
+    with open(bibliography) as f:
+        bib_lines = len(list(f))
+
+    tex_proc = subprocess.run(['git', 'diff', f'-U{tex_lines+1}', '--word-diff', '--color=always', revision, texfile],
+            check=True, capture_output=True)
+
+    bib_proc = subprocess.run(['git', 'diff', f'-U{bib_lines+1}', '--word-diff', '--color=always', revision, bibliography],
+            check=True, capture_output=True)
+
+    addition_re = re.compile('\033\\[32m\\{\\+(.*?)\\+\\}\033\\[m')
+    deletion_re = re.compile('\033\\[31m\\[-(.*?)\\-]\033\\[m')
+    csi_re = re.compile('\033\\[.*?m')
+    bibtex_entry_def_re = re.compile('@.*?{(.*?),')
+
+    bibliography_categories  = '\\DeclareBibliographyCategory{diff_new_entry}\n'
+    bibliography_categories += '\\DeclareBibliographyCategory{diff_deleted_entry}\n'
+    bibliography_categories += '\\AtEveryBibitem{\\ifcategory{diff_new_entry}{\\color{diffgreen}}{\\ifcategory{diff_deleted_entry}{\\color{diffred}}{\\color{black}}}}\n'
+
+    added_entries, removed_entries = [], []
+    for line in bib_proc.stdout.decode().splitlines():
+        if (match := addition_re.fullmatch(line.strip())):
+            if (entry_def := bibtex_entry_def_re.fullmatch(match.group(1))):
+                added_entries.append(entry_def.group(1))
+
+        if (match := deletion_re.fullmatch(line.strip())):
+            if (entry_def := bibtex_entry_def_re.fullmatch(match.group(1))):
+                removed_entries.append(entry_def.group(1))
+
+    if added_entries:
+        bibliography_categories += '\\addtocategory{diff_new_entry}{' + ','.join(added_entries) + '}\n'
+    if removed_entries:
+        bibliography_categories += '\\addtocategory{diff_deleted_entry}{' + ','.join(removed_entries) + '}\n'
+
+    content_started = False
+    document_started = False
+    for line in tex_proc.stdout.decode().splitlines():
+
+        if not content_started:
+            if '@@' in line:
+                content_started = True
+            continue
+
+        line = line.rstrip()
+        if document_started: # diff results in preamble
+            line = addition_re.sub(r'\\color{diffgreen}\1\\color{black}', line)
+            line = deletion_re.sub(r'\\color{diffred}\1\\color{black}', line)
+
+        else:
+            if '\\begin{document}' in line:
+                document_started = True
+                print(bibliography_categories)
+                print('\\definecolor{diffgreen}{HTML}{1e8449}')
+                print('\\definecolor{diffred}{HTML}{cb4335}')
+
+
+            line = addition_re.sub(r'\1', line)
+            line = deletion_re.sub(r'\1', line)
+
+        line = csi_re.sub('', line)
+        print(line)
+
+if __name__ == '__main__':
+    generate_git_tex_diff()
-- 
cgit