| 修订版 | df322ef3f0b99f289c308fe5f2cfaa4597225eeb (tree) |
|---|---|
| 时间 | 2013-04-18 19:35:55 |
| 作者 | hylom <hylom@hylo...> |
| Commiter | hylom |
add sakura_imgswap.py
sakura_imgswap.py (diff)| @@ -0,0 +1,115 @@ | ||
| 1 | +#!/usr/bin/env python | |
| 2 | +# -*- encoding: utf-8 -*- | |
| 3 | +# | |
| 4 | +# wp_imgswap.py : WordPress image tag swapper | |
| 5 | +# | |
| 6 | + | |
| 7 | +import sys | |
| 8 | +import codecs | |
| 9 | +import re | |
| 10 | +import os.path | |
| 11 | +import hashlib | |
| 12 | + | |
| 13 | +import htmltaglib | |
| 14 | +import deterfile | |
| 15 | +import getjpggeom | |
| 16 | + | |
| 17 | +usage = """usage: %s <target file> <output_file> <image_dir> [link_prefix] [image_width]""" % (os.path.basename(sys.argv[0]),) | |
| 18 | + | |
| 19 | +rex_imgtag = re.compile(r"""<img\s+src=["'](.*?)["'].*?>""") | |
| 20 | +rex_atag = re.compile(r"""<a\s+href=["'](.*?)["'].*?>""") | |
| 21 | +rex_alt = re.compile(r"""alt=["'](.*?)["']""") | |
| 22 | + | |
| 23 | +try: | |
| 24 | + in_f = codecs.open(sys.argv[1], "r", "utf_8" ) | |
| 25 | + out_f = codecs.open(sys.argv[2], "w", "utf_8" ) | |
| 26 | + image_dir = sys.argv[3] | |
| 27 | +except IndexError: | |
| 28 | + sys.exit(usage) | |
| 29 | + | |
| 30 | +try: | |
| 31 | + image_width = int(sys.argv[4]) | |
| 32 | +except IndexError: | |
| 33 | + image_width = 440 | |
| 34 | + | |
| 35 | + | |
| 36 | +def _get_png_geom(filepath): | |
| 37 | + s = filepath.split('.') | |
| 38 | + ext = s[-1] | |
| 39 | + if (ext == 'JPG') or (ext == 'jpg'): | |
| 40 | + (w, h) = getjpggeom.get_jpeg_geometory(filepath) | |
| 41 | + return (w, h) | |
| 42 | + else: | |
| 43 | + desc = deterfile.file(filepath) | |
| 44 | + | |
| 45 | + try: | |
| 46 | + m = re.match(r"([0-9]+)\s*x\s*([0-9]+)", desc[1]) | |
| 47 | + except IndexError: | |
| 48 | + err = ", ".join(desc) | |
| 49 | + raise Exception("deterfile error: %s, file: %s . " % (err,filepath)) | |
| 50 | + if m: | |
| 51 | + w = m.group(1) | |
| 52 | + h = m.group(2) | |
| 53 | + return (int(w), int(h)) | |
| 54 | + else: | |
| 55 | + return None | |
| 56 | + | |
| 57 | + | |
| 58 | +def replace_img_tag(line, tagstr, path): | |
| 59 | + if not os.path.isfile(path): | |
| 60 | + return line | |
| 61 | + | |
| 62 | + attrs = htmltaglib.parse_attributes(tagstr) | |
| 63 | + | |
| 64 | + if 'width' in attrs: | |
| 65 | + (w, h) = _get_png_geom(path) | |
| 66 | + if int(w) != image_width: | |
| 67 | + attrs['height'] = str(h * image_width / w) | |
| 68 | + attrs['width'] = str(image_width) | |
| 69 | + else: | |
| 70 | + attrs['height'] = str(h) | |
| 71 | + attrs['width'] = str(w) | |
| 72 | + | |
| 73 | + (root, ext) = os.path.splitext(os.path.basename(path)) | |
| 74 | + wp_image_url = '''%s%s-%sx%s%s''' % (image_dir, root, attrs['width'], attrs['height'], ext) | |
| 75 | + else: | |
| 76 | + wp_image_url = image_dir + os.path.basename(path) | |
| 77 | + | |
| 78 | + attrs['src'] = wp_image_url | |
| 79 | + # if tag has 'alt' attribute, use it | |
| 80 | + if rex_alt.search(tagstr): | |
| 81 | + alt_text = rex_alt.search(tagstr).group(1) | |
| 82 | + attrs['alt'] = alt_text | |
| 83 | + | |
| 84 | + new_tag_str = htmltaglib.build_tag('img', attrs) | |
| 85 | + return line.replace(tagstr, new_tag_str) | |
| 86 | + | |
| 87 | +def replace_a_tag(line, tagstr, path): | |
| 88 | + if not os.path.isfile(path): | |
| 89 | + return line | |
| 90 | + | |
| 91 | + attrs = htmltaglib.parse_attributes(tagstr) | |
| 92 | + wp_image_url = image_dir + os.path.basename(path) | |
| 93 | +# (root, ext) = os.path.splitext(os.path.basename(path)) | |
| 94 | +# wp_image_url = link_prefix + "/" + root | |
| 95 | + attrs['href'] = wp_image_url | |
| 96 | + new_tag_str = htmltaglib.build_tag('a', attrs) | |
| 97 | + | |
| 98 | + return line.replace(tagstr, new_tag_str) | |
| 99 | + | |
| 100 | +for line in in_f: | |
| 101 | + # proc for IMG tag | |
| 102 | + match = rex_imgtag.search(line) | |
| 103 | + if match: | |
| 104 | + tagstr = match.group(0) | |
| 105 | + path = match.group(1) | |
| 106 | + line = replace_img_tag(line, tagstr, path) | |
| 107 | + | |
| 108 | + #proc for A tag | |
| 109 | + match = rex_atag.search(line) | |
| 110 | + if match: | |
| 111 | + tagstr = match.group(0) | |
| 112 | + path = match.group(1) | |
| 113 | + line = replace_a_tag(line, tagstr, path) | |
| 114 | + | |
| 115 | + print >> out_f, line, |
Fork