pdf-auto-outline 0.1.6__tar.gz → 0.1.7__tar.gz

{pdf_auto_outline-0.1.6 → pdf_auto_outline-0.1.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: pdf-auto-outline
-Version: 0.1.6
+Version: 0.1.7
 Summary: Automatically generate and edit PDF table of contents / outline
 Author: Rossikos
 Author-email: Rossikos <216631970+rossikos@users.noreply.github.com>
@@ -91,7 +91,7 @@ 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 "%{file_path}"
+new_command _edit_toc pdfao "%{file_path}" -e
 ```
 
 The sioyek library and flag are optional; they allow logging to the status bar. This is more useful for ToC generation where you may want a progress bar. 

{pdf_auto_outline-0.1.6 → pdf_auto_outline-0.1.7}/README.md

@@ -74,7 +74,7 @@ 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 "%{file_path}"
+new_command _edit_toc pdfao "%{file_path}" -e
 ```
 
 The sioyek library and flag are optional; they allow logging to the status bar. This is more useful for ToC generation where you may want a progress bar. 

{pdf_auto_outline-0.1.6 → pdf_auto_outline-0.1.7}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "pdf-auto-outline"
-version = "0.1.6"
+version = "0.1.7"
 description = "Automatically generate and edit PDF table of contents / outline"
 readme = "README.md"
 authors = [

pdf_auto_outline-0.1.7/src/pdf_auto_outline/__init__.py

@@ -0,0 +1 @@
+__version__ = '0.1.7'

{pdf_auto_outline-0.1.6 → pdf_auto_outline-0.1.7}/src/pdf_auto_outline/main.py

@@ -160,7 +160,7 @@ def generate_txtfile(toc_entries, txtfile=get_tmpfile()):
                      TABLE OF CONTENTS OUTLINE
     4spaces/lvl text  |  pg#  |  {details dictionary} OR y-coord
 
-       Type 'C' as the first character of this file to cancel
+       Type '\\' as the first character of this file to cancel
     ============================================================
 
     """)
@@ -183,39 +183,56 @@ def generate_txtfile(toc_entries, txtfile=get_tmpfile()):
 
 def parse_txtfile(f, tablevel=2) -> list:
     toc_entries = []
-    if (c := f.read(1)) == 'C':
+    if (c := f.read(1)) == '\\':
         log('Outline not written')
         exit()
     elif c == '=':
-        lines = f.readlines()[7:]
-    else: lines = f.read()
+        for _ in range(7):
+            f.readline()
+    else: 
+        f.seek(0)
 
-    for i in lines:
+    for ln, i in enumerate(f):
         i = i.replace('\t', '    '*tablevel)
         lvl = (len(i) - len(i.lstrip())) // 4 + 1
         a = i.lstrip().split('  |  ')
-        if len(a) < 3:
+        # print(i)
+        if (l := len(a)) == 2:
             toc_entries.append(
                     [lvl, a[0], int(a[1])] 
             )
-        else:
+        elif l == 3:
             toc_entries.append(
                     [lvl, a[0], int(a[1]), eval(a[2])]
             )
+        else:
+            log(f'Error parsing line {ln+1}: {i}')
+            exit()
     
     f.close()
 
     return toc_entries
 
-def embed_toc(pdfpath, toc_entries, newfile=''):
+def embed_toc(pdfpath, toc_entries, newfile='', offset=0):
+    if offset != 0:
+        toc_entries = [[a, b, c + offset, *d] for a, b, c, *d in toc_entries]
     doc = pymupdf.open(pdfpath)
     doc.set_toc(toc_entries, collapse=2)
     if newfile:
         doc.save(newfile)
         log(f"toc written to '{newfile}'")
-    else:
+    elif doc.can_save_incrementally():
         doc.saveIncr()
         log(f"toc saved to '{pdfpath}'")
+    else:
+        log('cannot save to original; saving to new file...')
+        new_path = '_new.'.join(pdfpath.split('.'))
+        try: 
+            doc.save(new_path, garbage=4, deflate=True, use_objstms=True)
+        except KeyboardInterrupt as e:
+            log('Cancelled')
+            exit()
+        log(f"toc written to '{new_path}'")
 
 def get_toc_custom(doc) -> list:
     toc_entries = [[*i[:3], i[3].get('to')[1]] for i in doc.get_toc(False)]
@@ -238,8 +255,9 @@ def main():
     parser.add_argument('-se', '--superedit', action='store_true', help='edit pdf toc (more attibutes available)')
     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('-os', '--offset', type=int, metavar='<n>', help='toc page offset; use with infile', default=0)
     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 0.1.6')
+    parser.add_argument('--version', action='version', version='%(prog)s 0.1.7')
 
     args = parser.parse_args()
 
@@ -265,11 +283,11 @@ def main():
             f = generate_txtfile(get_toc_custom(doc))
         edit_txtfile(f)
         toc_entries = parse_txtfile(f, args.tablevel)
-        embed_toc(args.filename, toc_entries, args.out)
+        embed_toc(args.filename, toc_entries, args.out, args.offset)
         os.remove(f.name)
     elif args.infile:
         toc_entries = parse_txtfile(open(args.infile, encoding='utf-8'), args.tablevel)
-        embed_toc(args.filename, toc_entries, args.out)
+        embed_toc(args.filename, toc_entries, args.out, args.offset)
     else: # generate toc
         start = perf_counter()
         toc_entries = generate_toc_nnet(args.filename, args.multiprocess)
@@ -278,12 +296,12 @@ def main():
         log(f"finished in {end - start:<4.1f} s")
         toc_entries = align_toc_lvls(toc_entries)
         if args.straight:
-            embed_toc(args.filename, toc_entries, args.out)
+            embed_toc(args.filename, toc_entries, args.out, args.offset)
         else:
             f = generate_txtfile(toc_entries)
             edit_txtfile(f)
             toc_entries = parse_txtfile(f, args.tablevel)
-            embed_toc(args.filename, toc_entries, args.out)
+            embed_toc(args.filename, toc_entries, args.out, args.offset)
             os.remove(f.name)
 
     # if args.sioyek and not args.out:

pdf_auto_outline-0.1.7/src/pdf_auto_outline/test.py

@@ -0,0 +1,10 @@
+f = 'test.txt'
+ 
+with open(f) as f:
+    for _ in range(7):
+        f.readline()
+
+    for idi, i in enumerate(f):
+        print(idi, i)
+
+ 

pdf_auto_outline-0.1.7/src/pdf_auto_outline/test.txt

@@ -0,0 +1,13 @@
+============================================================
+                 TABLE OF CONTENTS OUTLINE
+4spaces/lvl text  |  pg#  |  {details dictionary} OR y-coord
+
+   Type '\\' as the first character of this file to cancel
+============================================================
+
+1
+2
+3
+4
+5
+6

pdf_auto_outline-0.1.6/src/pdf_auto_outline/__init__.py

@@ -1 +0,0 @@
-__version__ = '0.1.6'

pdf_auto_outline-0.1.6/src/pdf_auto_outline/test.py

@@ -1,4 +0,0 @@
-txt = "aa\nbb\ncc"
-
-for i in txt:
-    print(i)