pdf-auto-outline 0.1.5__tar.gz → 0.1.7__tar.gz

{pdf_auto_outline-0.1.5 → pdf_auto_outline-0.1.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: pdf-auto-outline
-Version: 0.1.5
+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>
@@ -82,7 +82,6 @@ The optional part can be one of:
 ```
   |  None                  	same as not including it   
   |  241.2					y-ordinate 
-  |  Point(72.0, 363.9)		x and y coords
   |  {<dictionary>}			dictionary with more attributes for the ToC entry
 ```
 
@@ -92,8 +91,10 @@ 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}" --sioyek path/to/sioyek -e
+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. 
+
 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-0.1.5 → pdf_auto_outline-0.1.7}/README.md

@@ -65,7 +65,6 @@ The optional part can be one of:
 ```
   |  None                  	same as not including it   
   |  241.2					y-ordinate 
-  |  Point(72.0, 363.9)		x and y coords
   |  {<dictionary>}			dictionary with more attributes for the ToC entry
 ```
 
@@ -75,8 +74,10 @@ 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}" --sioyek path/to/sioyek -e
+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. 
+
 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-0.1.5 → pdf_auto_outline-0.1.7}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "pdf-auto-outline"
-version = "0.1.5"
+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-0.1.5 → pdf_auto_outline-0.1.7}/src/pdf_auto_outline/main.py

@@ -1,10 +1,10 @@
 import pymupdf.layout
-from pymupdf import Point
 from time import perf_counter
 from multiprocessing import Pool
 import os
 import subprocess
 import argparse
+import tempfile
 
 SIOYEK = None
 
@@ -148,14 +148,19 @@ def align_toc_lvls(toc_entries: list) -> list:
 
     return toc_entries
 
-def generate_txtfile(toc_entries, txtfile='outline.txt') -> str:
+def get_tmpfile():
+    return tempfile.NamedTemporaryFile(
+            mode='w+', encoding='utf-8', delete=False, suffix='.txt'
+            )
+
+def generate_txtfile(toc_entries, txtfile=get_tmpfile()):
     import textwrap
     txt = textwrap.dedent("""\
     ============================================================
                      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
     ============================================================
 
     """)
@@ -168,60 +173,77 @@ def generate_txtfile(toc_entries, txtfile='outline.txt') -> str:
         txt += '\n'.join(f"{' '*4 * (i[0] - 1)}{i[1]}  |  {i[2]}"
                         for i in toc_entries)
 
-    with open(txtfile, 'w', encoding='utf-8') as f:
-        f.write(txt)
+    txtfile.write(txt)
+    txtfile.flush()
+    txtfile.seek(0)
 
     return txtfile
 
 
-def parse_txtfile(txtfile='outline.txt', tablevel=2) -> list:
+
+def parse_txtfile(f, tablevel=2) -> list:
     toc_entries = []
-    with open(txtfile) as f:
-        if (c := f.read(1)) == 'C':
-            log('Outline not written')
+    if (c := f.read(1)) == '\\':
+        log('Outline not written')
+        exit()
+    elif c == '=':
+        for _ in range(7):
+            f.readline()
+    else: 
+        f.seek(0)
+
+    for ln, i in enumerate(f):
+        i = i.replace('\t', '    '*tablevel)
+        lvl = (len(i) - len(i.lstrip())) // 4 + 1
+        a = i.lstrip().split('  |  ')
+        # print(i)
+        if (l := len(a)) == 2:
+            toc_entries.append(
+                    [lvl, a[0], int(a[1])] 
+            )
+        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()
-        elif c == '=':
-            lines = f.readlines()[7:]
-        else: lines = f.read()
-
-        for i in lines:
-            i = i.replace('\t', '    '*tablevel)
-            lvl = (len(i) - len(i.lstrip())) // 4 + 1
-            a = i.lstrip().split('  |  ')
-            if len(a) < 3:
-                toc_entries.append(
-                        [lvl, a[0], int(a[1])] 
-                )
-            else:
-                toc_entries.append(
-                        [lvl, a[0], int(a[1]), eval(a[2])]
-                )
+    
+    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)]
     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', 'start' if os.name == 'nt' else 'xdg-open')
-    name = os.name
-    if name == 'nt':
-        subprocess.run(['start', '/WAIT', txtfile], shell=True)
+def edit_txtfile(f):
+    if os.name == 'nt':
+        subprocess.run(['start', '/WAIT', f.name], shell=True)
     else: # name == 'posix':
         editor = os.environ.get('EDITOR', 'vi')  
-        subprocess.run([editor, txtfile])
+        subprocess.run([editor, f.name])
 
 def main():
     parser = argparse.ArgumentParser(prog='pdfao')
@@ -233,20 +255,21 @@ 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 0.1.5')
+    parser.add_argument('--version', action='version', version='%(prog)s 0.1.7')
 
     args = parser.parse_args()
 
+    if args.out:
+        args.out = os.path.join(
+                os.path.dirname(args.filename),
+                args.out)
+
     if args.sioyek:
         from sioyek.sioyek import Sioyek
         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]
@@ -255,15 +278,16 @@ def main():
     if args.edit or args.superedit:
         doc = pymupdf.Document(args.filename)
         if args.superedit:
-            generate_txtfile(doc.get_toc(False))
+            f = 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)
+            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, args.offset)
+        os.remove(f.name)
     elif args.infile:
-        toc_entries = parse_txtfile(args.infile, args.tablevel)
-        embed_toc(args.filename, toc_entries, args.out)
+        toc_entries = parse_txtfile(open(args.infile, encoding='utf-8'), args.tablevel)
+        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)
@@ -272,12 +296,13 @@ 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:
-            generate_txtfile(toc_entries)
-            edit_txtfile()
-            toc_entries = parse_txtfile(tablevel=args.tablevel)
-            embed_toc(args.filename, toc_entries, args.out)
+            f = generate_txtfile(toc_entries)
+            edit_txtfile(f)
+            toc_entries = parse_txtfile(f, args.tablevel)
+            embed_toc(args.filename, toc_entries, args.out, args.offset)
+            os.remove(f.name)
 
     # if args.sioyek and not args.out:
     #     to_hash = get_md5_hash(args.filename)

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-0.1.5/src/pdf_auto_outline/__init__.py

@@ -1 +0,0 @@
-__version__ = '0.1.5'