pdf-auto-outline 0.1.0__py3-none-any.whl → 0.1.2__py3-none-any.whl

pdf_auto_outline/__init__.py

@@ -1 +1 @@
-__version__ = '0.1.0'
+__version__ = '0.1.2'

pdf_auto_outline/__main__.py

@@ -1,4 +1,4 @@
-from .main import main
+from pdf_auto_outline.main import main
 
 if __name__ == '__main__':
     main()

pdf_auto_outline/main.py

@@ -96,77 +96,56 @@ def generate_toc_nnet(pdfpath, worker_cnt=3) -> list:
         log('\nCancelled')
         exit()
 
-    log('')
 
     return [j for i in pg_nums for j in results[i]]
 
 def align_toc_lvls(toc_entries: list) -> list:
-    # TODO: fix this spaghetti
     import re
-    def act(lvl, current, prev): # cur prev expected lvl
-        # if current == prev - 1: # current is parent
-        if current == prev[0]: # current is sibling
-            return lvl
-        elif current == 'p5':
-            return lvl + 1
-        elif e[current] < prev[1]: # current is parent
-            return e[current]
-            # return max(1, lvl - 1)
+    def act(current): # cur prev expected lvl
+        if current == d['prev_name']: # current is sibling
+            pass
+        elif current == 'p5': # current is figure/table type
+            d['lvl'] += 1
+        elif e[current] < d['prev_lvl']: # current is parent
+            d['lvl'] = e[current]
         else: # e[current] > prev[1]: # current is child
-            e[current] = min(lvl + 1, e[current])
-            return min(lvl + 1, e[current])
-        # else: #e[current] == prev: # current is sibling
-        #     return lvl
+            e[current] = min(d['lvl'] + 1, e[current])
+            d['lvl'] = min(d['lvl'] + 1, e[current])
 
-    p1 = re.compile(r'^[A-Z\d]')
-    p2 = re.compile(r'^(Contents)|(Chapter)|(Appendix)|(Index)|(Bibliograph)|(Preface)')
-    p3 = re.compile(r'^([IVXC\d])+\.[IVXC\d]\.? \w')
-    p4 = re.compile(r'^([AIVXC\d]+\.){2}[IVXC\d]\.? \w')
-    p5 = re.compile(r'^(Fig(ure)?\.?)|(Table\.? [\dIVXC]+)')
-    p6 = re.compile(r'''\d?\s?(Introduction)|((Materials and )?Methods)|(Results)|
-                    (Discussion)|(References)|(Summary)|(Conclusion)|(Acknowledgements)
-                    ''', re.IGNORECASE)
-    p7 = re.compile(r'^\d?\s?[A-Z ]{2,}') 
+        d['prev_name'] = current
+        d['prev_lvl'] = e[current]
+        toc_entries[i-d['removed']][0] = d['lvl']
 
+    p1 = re.compile(r'^[A-Z\d]')
+    patterns = (
+        re.compile(r'^(Contents)|(Chapter)|(Appendix)|(Index)|(Bibliography)|(Preface)'),
+        re.compile(r'^([IVXC\d])+\.[IVXC\d]\.? \w'),
+        re.compile(r'^([AIVXC\d]+\.){2}[IVXC\d]\.? \w'),
+        re.compile(r'^(Fig(ure)?\.?)|(Table\.? [\dIVXC]+)', re.IGNORECASE),
+        re.compile(r'''\d?\s?(Introduction)|((Materials? and )?Methods)|(Results)|
+                        (Discussion)|(References)|(Summary)|(Conclusion)|(Acknowledgements)
+                        ''', re.IGNORECASE),
+        re.compile(r'^\d?\s?[A-Z ]{2,}'),
+    )
+
+    # expected nesting levels
     e = {'p1': 1, 'p2': 1, 'p3': 2, 'p4': 3, 'p5': 5, 'p6': 1, 'p7': 1, 'l': 2,}
+    # line status
+    d = {'lvl': 1, 'prev_name': 'p1', 'prev_lvl': 1, 'titles': set(), 'removed': 0}
 
     log('aligning levels..')
-    lvl, prev, titles, removed = 1, ('p1', 1), set(), 0
 
     for i in range(1, len(toc_entries)):
-        title = toc_entries[i-removed][1]
-        if (not p1.match(title)) or len(title) < 4 or title in titles: #skip
-            toc_entries.pop(i-removed)
-            removed += 1
-        elif p2.match(title):
-            lvl = act(lvl, 'p2', prev)
-            toc_entries[i-removed][0] = lvl
-            prev = ('p2', e['p2'])
-        elif p7.match(title):
-            lvl = act(lvl, 'p7', prev)
-            toc_entries[i-removed][0] = lvl
-            prev = ('p7', e['p7'])
-        elif p6.match(title):
-            lvl = act(lvl, 'p6', prev)
-            toc_entries[i-removed][0] = lvl
-            prev = ('p6', e['p6'])
-        elif p3.match(title):
-            lvl = act(lvl, 'p3', prev)
-            toc_entries[i-removed][0] = lvl
-            prev = ('p3', e['p3'])
-        elif p4.match(title):
-            lvl = act(lvl, 'p4', prev)
-            toc_entries[i-removed][0] = lvl
-            prev = ('p4', e['p4'])
-        elif p5.match(title):
-            lvl = act(lvl, 'p5', prev)
-            toc_entries[i-removed][0] = lvl
-            prev = ('p5', e['p5'])
+        title = toc_entries[i-d['removed']][1]
+        if (not p1.match(title)) or len(title) < 4 or title in d['titles']: #skip
+            toc_entries.pop(i-d['removed'])
+            d['removed'] += 1
+        elif (name := next((idi for idi, i in enumerate(patterns) if i.match(title)), None)):
+            act(f'p{name+2}')
         else:
-            titles.add(title)
-            lvl = act(lvl, 'l', prev)
-            toc_entries[i-removed][0] = lvl
-            prev = ('l', e['l'])
+            d['titles'].add(title)
+            act('l')
+
     return toc_entries
 
 def generate_txtfile(toc_entries, txtfile='outline.txt') -> str:
@@ -214,7 +193,6 @@ def parse_txtfile(txtfile='outline.txt', tablevel=2) -> list:
     return toc_entries
 
 def embed_toc(pdfpath, toc_entries, newfile=''):
-    print(len(toc_entries))
     doc = pymupdf.open(pdfpath)
     doc.set_toc(toc_entries, collapse=2)
     if newfile:
@@ -224,10 +202,13 @@ def embed_toc(pdfpath, toc_entries, newfile=''):
         doc.saveIncr()
         log(f"toc saved to '{pdfpath}'")
 
-
+def get_toc_custom(doc) -> list:
+    toc_entries = [[*i[:3], i[3].get('to')] for i in doc.get_toc(False)]
+    return toc_entries
 
 def edit_txtfile(txtfile='outline.txt'):
-    editor = os.environ.get('EDITOR', 'notepad' if os.name == 'nt' else 'vi')
+    # editor = os.environ.get('EDITOR', 'notepad' if os.name == 'nt' else 'vi')
+    editor = os.environ.get('EDITOR', 'start' if os.name == 'nt' else 'xdg-open')
     subprocess.run([editor, txtfile])
 
 def main():
@@ -241,14 +222,19 @@ def main():
     parser.add_argument('-i', '--infile', type=str, metavar='<file>', help='write toc from file to pdf')
     parser.add_argument('-t', '--tablevel', type=int, metavar='<n>', help='tab = n toc nesting levels (default 2)', default=2)
     parser.add_argument('--sioyek', type=str, metavar='<path>', help='for users of the Sioyek pdf viewer')
-    parser.add_argument('--version', action='version', version='%(prog)s 0.1.0')
+    parser.add_argument('--version', action='version', version='%(prog)s 0.1.2')
 
     args = parser.parse_args()
 
     if args.sioyek:
         from sioyek.sioyek import Sioyek
-        sioyek_path = args.sioyek[0]
-        SIOYEK = Sioyek(sioyek_path)
+        global SIOYEK
+        SIOYEK = Sioyek(args.sioyek)
+        if args.out:
+            args.out = os.path.join(
+                    os.path.dirname(args.filename),
+                    args.out
+                )
         # local_db = args.sioyek[1]
         # shared_db = args.sioyek[2]
         # pdf_path = args.sioyek[3]
@@ -256,7 +242,10 @@ def main():
 
     if args.edit or args.superedit:
         doc = pymupdf.Document(args.filename)
-        generate_txtfile(doc.get_toc(not args.superedit))
+        if args.superedit:
+            generate_txtfile(doc.get_toc(False))
+        else:
+            generate_txtfile(get_toc_custom(doc))
         edit_txtfile()
         toc_entries = parse_txtfile(tablevel=args.tablevel)
         embed_toc(args.filename, toc_entries, args.out)
@@ -267,6 +256,7 @@ def main():
         start = perf_counter()
         toc_entries = generate_toc_nnet(args.filename, args.multiprocess)
         end = perf_counter()
+        log('')
         log(f"finished in {end - start:<4.1f} s")
         toc_entries = align_toc_lvls(toc_entries)
         if args.straight:

pdf_auto_outline-0.1.2.dist-info/METADATA

@@ -0,0 +1,73 @@
+Metadata-Version: 2.3
+Name: pdf-auto-outline
+Version: 0.1.2
+Summary: Automatically generate and edit PDF table of contents / outline
+Author: Rossikos
+Author-email: Rossikos <216631970+rossikos@users.noreply.github.com>
+Classifier: Programming Language :: Python :: 3
+Classifier: License :: OSI Approved :: GNU General Public License v3 (GPLv3)
+Classifier: Operating System :: OS Independent
+Requires-Dist: pymupdf-layout>=1.26.6
+Requires-Dist: sioyek ; extra == 'sioyek'
+Requires-Python: >=3.13
+Project-URL: Bug Tracker, https://github.com/rossikos/pdf-auto-outline/issues
+Project-URL: Homepage, https://github.com/rossikos/pdf-auto-outline
+Provides-Extra: sioyek
+Description-Content-Type: text/markdown
+
+# PDF Auto Outline
+
+Automatically generate and embed a table of contents or outline in a PDF.
+
+Install: `python -m pip install pdf-auto-outline`
+
+Suggestions and contributions are welcome.
+
+## Usage
+
+```
+usage: pdfao [-h] [-s] [-o <path>] [-mp <n>] [-e] [-se] [-i <file>] [-t <n>] [--sioyek <path>] [--version] filename
+
+positional arguments:
+  filename              input pdf
+
+options:
+  -h, --help            show this help message and exit
+  -s, --straight        write toc straight to pdf; skip editing
+  -o, --out <path>      write changes to new pdf
+  -mp, --multiprocess <n>
+                        spread job over n processes (faster on Linux)
+  -e, --edit            edit pdf toc
+  -se, --superedit      edit pdf toc (more attibutes available)
+  -i, --infile <file>   write toc from file to pdf
+  -t, --tablevel <n>    tab = n toc nesting levels (default 2)
+  --sioyek <path>       for users of the Sioyek pdf viewer
+  --version             show program's version number and exit
+```
+
+### Examples
+
+Generate toc and edit before saving:
+`pdfao paper.pdf`
+
+Generate and save to new pdf:
+`pdfao paper.pdf -o new.pdf`
+
+Edit exiting pdf toc:
+`pdfao paper.pdf -e`
+
+A save toc to new pdf from file:
+`pdfao paper.pdf -o new.pdf -i outline.txt`
+
+
+## For Sioyek Users
+
+Example commands; add to `prefs_user.config`.
+
+```
+new_command _gen_toc pdfao "%{file_path}" --sioyek path/to/sioyek -mp 4
+new_command _edit_toc pdfao path/to/pdfao.py "%{file_path}" --sioyek path/to/sioyek -e
+```
+
+If you don't wish to install from PyPI, download source and use `python3 -m path/to/src/pdf_auto_outline` in place of `pdfao`.
+

pdf_auto_outline-0.1.2.dist-info/RECORD

@@ -0,0 +1,7 @@
+pdf_auto_outline/__init__.py,sha256=mdp2CftfqYbdKtP-eWv1z7rAUycYv6X1ntXSMUf8Kss,22
+pdf_auto_outline/__main__.py,sha256=mRKsAFeG5R17vTYubIKregAve4vnKc-nk7jY3tcK4wI,78
+pdf_auto_outline/main.py,sha256=oZJMNGDBLJNGWmjGtDdNoFXsPtSS7Km7uPdwjHXPA00,9638
+pdf_auto_outline-0.1.2.dist-info/WHEEL,sha256=5w2T7AS2mz1-rW9CNagNYWRCaB0iQqBMYLwKdlgiR4Q,78
+pdf_auto_outline-0.1.2.dist-info/entry_points.txt,sha256=HBvhmxJs8hHqbbpJmVTbBH3xy19Hk655O_ySwFC_53w,100
+pdf_auto_outline-0.1.2.dist-info/METADATA,sha256=GyMWMb7y2SDwvxUEEMpVpz4Fxfe3_GwGJ-sdXHxAFZg,2316
+pdf_auto_outline-0.1.2.dist-info/RECORD,,

pdf_auto_outline-0.1.0.dist-info/METADATA

@@ -1,46 +0,0 @@
-Metadata-Version: 2.3
-Name: pdf-auto-outline
-Version: 0.1.0
-Summary: Automatically generate and edit PDF table of contents / outline
-Author: Rossikos
-Author-email: Rossikos <216631970+rossikos@users.noreply.github.com>
-Requires-Dist: pymupdf-layout>=1.26.6
-Requires-Python: >=3.13
-Description-Content-Type: text/markdown
-
-# PDF Auto Outline
-
-A simple python program to automatically generate and embed a table of contents or outline in a PDF.
-
-## Usage
-
-```
-usage: pdfao [-h] [-s] [-o <path>] [-mp <n>] [-e] [-se] [-i <file>] [-t <n>] [--sioyek <path>] [--version] filename
-
-positional arguments:
-  filename              input pdf
-
-options:
-  -h, --help            show this help message and exit
-  -s, --straight        write toc straight to pdf; skip editing
-  -o, --out <path>      write changes to new pdf
-  -mp, --multiprocess <n>
-                        spread job over n processes (faster on linux)
-  -e, --edit            edit pdf toc
-  -se, --superedit      edit pdf toc (more attibutes available)
-  -i, --infile <file>   write toc from file to pdf
-  -t, --tablevel <n>    tab = n toc nesting levels (default 2)
-  --sioyek <path>       for users of the Sioyek pdf viewer
-  --version             show program's version number and exit
-```
-
-## For Sioyek Users
-
-Example commands; add to prefs_user.config.
-
-```
-new_command _gen_toc python3 path/to/pdfao.py "%{file_path}" --sioyek path/to/sioyek -mp 4
-new_command _edit_toc python3 path/to/pdfao.py "%{file_path}" --sioyek path/to/sioyek -e
-```
-
-

pdf_auto_outline-0.1.0.dist-info/RECORD

@@ -1,7 +0,0 @@
-pdf_auto_outline/__init__.py,sha256=IMjkMO3twhQzluVTo8Z6rE7Eg-9U79_LGKMcsWLKBkY,22
-pdf_auto_outline/__main__.py,sha256=7tzuGbeA5JiJWE_g9pzlcTXSsKlR-iEXNEbdYd4jZMs,62
-pdf_auto_outline/main.py,sha256=KkLEIGCndRql55jjvAb8Y-onmkPoQwdExHcuNG3MPYw,9977
-pdf_auto_outline-0.1.0.dist-info/WHEEL,sha256=5w2T7AS2mz1-rW9CNagNYWRCaB0iQqBMYLwKdlgiR4Q,78
-pdf_auto_outline-0.1.0.dist-info/entry_points.txt,sha256=HBvhmxJs8hHqbbpJmVTbBH3xy19Hk655O_ySwFC_53w,100
-pdf_auto_outline-0.1.0.dist-info/METADATA,sha256=ZeQBy-6lWQbqhTmt2375o5c-CQymnOffWyFeL7vi3bY,1504
-pdf_auto_outline-0.1.0.dist-info/RECORD,,