Coverage for gws-app/gws/lib/pdf/__init__.py: 0%

1import pypdf 

2 

3 

4def overlay(a_path: str, b_path: str, out_path: str) -> str: 

5 """Overlay two pdfs page-wise. 

6 

7 Args: 

8 a_path: Path to pdf a. 

9 b_path: Path to pdf b, which will be placed on top. 

10 out_path: Path to the output pdf. 

11 

12 Returns: 

13 Path to the output pdf. 

14 """ 

15 

16 fa = open(a_path, 'rb') 

17 fb = open(b_path, 'rb') 

18 

19 ra = pypdf.PdfReader(fa) 

20 rb = pypdf.PdfReader(fb) 

21 

22 w = pypdf.PdfWriter() 

23 

24 for n, page in enumerate(ra.pages): 

25 other = None 

26 try: 

27 other = rb.pages[n] 

28 except IndexError: 

29 pass 

30 if other: 

31 # https://github.com/py-pdf/pypdf/issues/2139 

32 page.transfer_rotation_to_content() 

33 page.merge_page(other) 

34 w.add_page(page) 

35 

36 with open(out_path, 'wb') as out_fp: 

37 w.write(out_fp) 

38 

39 fa.close() 

40 fb.close() 

41 

42 return out_path 

43 

44 

45def concat(paths: [str], out_path: str) -> str: 

46 """Concatenates multiple pdfs into one. 

47 The order of the pages are the same as the order of the paths in the list. 

48 

49 Args: 

50 paths: Paths to the pdfs. 

51 out_path: Path to the output pdf. 

52 

53 Returns: 

54 Path to the concatenated pdf. 

55 """ 

56 

57 # only one path given - just return it 

58 if len(paths) == 1: 

59 return paths[0] 

60 

61 # NB: readers must be kept around until the writer is done 

62 

63 files = [open(p, 'rb') for p in paths] 

64 readers = [pypdf.PdfReader(fp) for fp in files] 

65 

66 w = pypdf.PdfWriter() 

67 

68 for r in readers: 

69 w.append_pages_from_reader(r) 

70 

71 with open(out_path, 'wb') as out_fp: 

72 w.write(out_fp) 

73 

74 for fp in files: 

75 fp.close() 

76 

77 return out_path 

78 

79 

80def page_count(path: str) -> int: 

81 """Returns the amount of pages for a given pdf. 

82 

83 Args: 

84 path: Path to the pdf. 

85 

86 Returns: 

87 Amount of pages in the pdf. 

88 """ 

89 with open(path, 'rb') as fp: 

90 r = pypdf.PdfReader(fp) 

91 return len(r.pages) 

92 

93# def to_image(in_path, out_path, size, mime): 

94# if mime == gws.lib.mime.PNG: 

95# device = 'png16m' 

96# elif mime == gws.lib.mime.JPEG: 

97# device = 'jpeg' 

98# else: 

99# raise ValueError(f'uknown format {format!r}') 

100# 

101# cmd = [ 

102# 'gs', 

103# '-q', 

104# f'-dNOPAUSE', 

105# f'-dBATCH', 

106# f'-dDEVICEWIDTHPOINTS={size[0]}', 

107# f'-dDEVICEHEIGHTPOINTS={size[1]}', 

108# f'-dPDFFitPage=true', 

109# f'-sDEVICE={device}', 

110# f'-dTextAlphaBits=4', 

111# f'-dGraphicsAlphaBits=4', 

112# f'-sOutputFile={out_path}', 

113# in_path, 

114# ] 

115# 

116# gws.log.debug(repr(cmd)) 

117# gws.lib.osx.run(cmd) 

118# 

119# return out_path