|
| 1 | +import argparse |
| 2 | +import glob |
| 3 | +import logging |
| 4 | +import multiprocessing |
| 5 | +import os |
| 6 | +import pathlib |
| 7 | +import re |
| 8 | +import subprocess |
| 9 | +import sys |
| 10 | +import tempfile |
| 11 | + |
| 12 | + |
| 13 | +""" |
| 14 | +This script does two things that improve the website organization. |
| 15 | +
|
| 16 | +First, we used to host in the root of the webpage, but have now moved to |
| 17 | +``/stable/``. We do not want obsolete links to link to nothing (or that has |
| 18 | +been our policy), so we currently just keep the old version at the top level. |
| 19 | +Here, instead, we either softlink to the newest version, or replace the file by |
| 20 | +an html refresh redirect. |
| 21 | +
|
| 22 | +Second, it changes the canonical link in each html file to the newest version |
| 23 | +found of the html file (including stable if its in the latest version.) |
| 24 | +
|
| 25 | +This script takes a while, and is destructive, so should probably be run on a |
| 26 | +branch and pushed as a PR so it can easily be reverted. |
| 27 | +""" |
| 28 | + |
| 29 | +_log = logging.getLogger('make_redirect_links') |
| 30 | + |
| 31 | + |
| 32 | +tocheck = ['stable'] + [f'{major}.{minor}.{micro}' |
| 33 | + for major in range(6, -1, -1) |
| 34 | + for minor in range(6, -1, -1) |
| 35 | + for micro in range(6, -1, -1)] |
| 36 | + |
| 37 | +toignore = tocheck + ['mpl-probscale', 'mpl_examples', 'mpl_toolkits', |
| 38 | + '_webpageutils', 'xkcd', 'sitemap.xml', |
| 39 | + 'robots.txt', 'CNAME', '.git'] |
| 40 | + |
| 41 | +logging.basicConfig(level=logging.DEBUG) |
| 42 | + |
| 43 | + |
| 44 | +def findlast(fname, tocheck): |
| 45 | + """ |
| 46 | + Check the directories listed in ``tocheck`` to see if they have |
| 47 | + ``fname`` in them. Return the first one found, or None |
| 48 | + """ |
| 49 | + p = pathlib.Path(fname) |
| 50 | + for t in tocheck: |
| 51 | + pnew = pathlib.Path(t, p) |
| 52 | + if pnew.exists(): |
| 53 | + return t |
| 54 | + else: |
| 55 | + return None |
| 56 | + |
| 57 | +html_redirect = """ |
| 58 | +<!DOCTYPE HTML> |
| 59 | +<html lang="en"> |
| 60 | + <head> |
| 61 | + <meta charset="utf-8"> |
| 62 | + <meta http-equiv="refresh" content="0;url=https://matplotlib.org%s" /> |
| 63 | + <link rel="canonical" href="https://matplotlib.org%s" /> |
| 64 | + </head> |
| 65 | + <body> |
| 66 | + <h1> |
| 67 | + The page been moved to <a href="https://matplotlib.org%s"</a> |
| 68 | + </h1> |
| 69 | + </body> |
| 70 | +</html> |
| 71 | +""" |
| 72 | + |
| 73 | + |
| 74 | +def do_links(root0): |
| 75 | + """ |
| 76 | + Either soft link a file at the top level to its newest position, |
| 77 | + or make an html redirect if it is an html file. |
| 78 | + """ |
| 79 | + _log.info(f'Doing links on {root0}') |
| 80 | + for root, dirs, files in os.walk(root0): |
| 81 | + for name in files: |
| 82 | + fullname = os.path.join(root, name) |
| 83 | + last = findlast(fullname, tocheck) |
| 84 | + _log.debug(f'Checking: {fullname} found {last}') |
| 85 | + if last is not None: |
| 86 | + os.remove(fullname) |
| 87 | + if name.endswith(('.htm', '.html')): |
| 88 | + # make an html redirect. |
| 89 | + _log.info(f'Rewriting HTML: {fullname} in {last}') |
| 90 | + with open(fullname, 'w') as fout: |
| 91 | + oldname = '/' + os.path.join(last, fullname) |
| 92 | + st = html_redirect % (oldname, oldname, oldname) |
| 93 | + fout.write(st) |
| 94 | + else: |
| 95 | + # soft link |
| 96 | + # Need to do these relative to where the link is |
| 97 | + # so if it is a level down `ln -s ../3.1.1/boo/who boo/who` |
| 98 | + last = os.path.join('..', last) |
| 99 | + depth = root.count('/') |
| 100 | + for i in range(depth): |
| 101 | + last = os.path.join('..', last) |
| 102 | + oldname = os.path.join(last, fullname) |
| 103 | + _log.info(f'Linking {fullname} to {oldname}') |
| 104 | + os.symlink(oldname, fullname) |
| 105 | + for d in dirs: |
| 106 | + do_links(d) |
| 107 | + |
| 108 | + |
| 109 | +def do_canonicals(dname): |
| 110 | + """ |
| 111 | + For each html file in the versioned docs, make the canonical link point |
| 112 | + to the newest version. |
| 113 | + """ |
| 114 | + _log.debug(f'Walking {dname}') |
| 115 | + for root, dirs, files in os.walk(dname): |
| 116 | + for name in files: |
| 117 | + fullname = os.path.join(root, name) |
| 118 | + p = pathlib.Path(fullname) |
| 119 | + _log.debug(f'Checking {fullname}') |
| 120 | + if name.endswith(('.htm', '.html')): |
| 121 | + basename = pathlib.Path(*p.parts[1:]) |
| 122 | + last = findlast(basename, tocheck) |
| 123 | + if last is not None: |
| 124 | + update_canonical(fullname, last) |
| 125 | + |
| 126 | + for d in dirs: |
| 127 | + _log.info(f'DIR: {d}') |
| 128 | + do_canonicals(os.path.join(dname,d)) |
| 129 | + |
| 130 | + |
| 131 | +def update_canonical(fullname, last): |
| 132 | + """ |
| 133 | + Change the canonical link in *fullname* to the same link in the |
| 134 | + version given by *last*. We do this with a regexp to prevent |
| 135 | + removing any other content on a line that has the canonical link. |
| 136 | +
|
| 137 | + Note that if for some reason there are more than one canonical link |
| 138 | + this will change all of them. |
| 139 | + """ |
| 140 | + p = pathlib.Path(fullname) |
| 141 | + pre = 'https://matplotlib.org/' |
| 142 | + pnew = pathlib.Path(last, *p.parts[1:]) |
| 143 | + newcanon = f'{pre+str(pnew)}' |
| 144 | + _log.info(f'{p} to {pre+str(pnew)}') |
| 145 | + with tempfile.NamedTemporaryFile(delete=False) as fout: |
| 146 | + with open(fullname, 'rb') as fin: |
| 147 | + for line in fin: |
| 148 | + if b'<link rel="canonical"' in line: |
| 149 | + new = bytes(f'<link rel="canonical" href="{newcanon}"', |
| 150 | + encoding='utf-8') |
| 151 | + ll = re.sub(b'<link rel="canonical" href=".*"', new, |
| 152 | + line) |
| 153 | + _log.debug(f'new {line}->{ll}') |
| 154 | + fout.write(ll) |
| 155 | + else: |
| 156 | + fout.write(line) |
| 157 | + os.rename(fout.name, fullname) |
| 158 | + |
| 159 | + |
| 160 | +if __name__ == "__main__": |
| 161 | + |
| 162 | + parser = argparse.ArgumentParser(description='Optional app description') |
| 163 | + |
| 164 | + parser.add_argument('--np', type=int, help='Number of processors to use') |
| 165 | + parser.add_argument('--no_canonicals', help='do not do canonical links', |
| 166 | + action="store_true") |
| 167 | + parser.add_argument('--no_redirects', help='do not do redirects links', |
| 168 | + action="store_true") |
| 169 | + |
| 170 | + args = parser.parse_args() |
| 171 | + if args.np: |
| 172 | + np = args.np |
| 173 | + else: |
| 174 | + np = None |
| 175 | + |
| 176 | + # html redirect or soft link most things in the top-level directory that |
| 177 | + # are not other modules or versioned docs. |
| 178 | + if not args.no_redirects: |
| 179 | + for entry in os.scandir('./'): |
| 180 | + if not (entry.name in toignore): |
| 181 | + if entry.is_dir(): |
| 182 | + do_links(entry.name) |
| 183 | + elif entry.name.endswith(('.htm', '.html')): |
| 184 | + fullname = entry.name |
| 185 | + last = findlast(fullname, tocheck) |
| 186 | + _log.debug(f'Checking: {fullname} found {last}') |
| 187 | + if last is not None: |
| 188 | + os.remove('./'+fullname) |
| 189 | + _log.info(f'Rewriting HTML: {fullname} in {last}') |
| 190 | + with open(fullname, 'w') as fout: |
| 191 | + oldname = '/' + os.path.join(last, fullname) |
| 192 | + st = html_redirect % (oldname, oldname, oldname) |
| 193 | + fout.write(st) |
| 194 | + _log.info('Done links and redirects') |
| 195 | + |
| 196 | + # change the canonical url for all html to the newest version in the docs: |
| 197 | + if not args.no_canonicals: |
| 198 | + if np is not None: |
| 199 | + with multiprocessing.Pool(np) as pool: |
| 200 | + pool.map(do_canonicals, tocheck[1:]) |
| 201 | + else: |
| 202 | + for t in tocheck[1:]: |
| 203 | + do_canonicals(t) |
0 commit comments