oraround the file's text and a "ReadmeName README.html" in the .htaccess file. For less-important cases, rename to "_README.txt" and let the user click if they really wish to view; a script can easily automate this: see learning-python.com/fix-readmes.py for an example. UPDATE, Apr-2018: server breakage #1 Oddly, this story differs at a new server to which this site was recently moved. On the new server (only), auto-index pages list README.txt files, but do not display their content inline, even if named in ReadmeName directives. Hence, fix-readmes.py is not required, and if used must be accommodated by IndexIgnore to hide any _README*s. Alas, Apache's wildly implicit design yields radically variable servers! In retrospect, README files might also have been excluded from showcode by enhanced rewrite-rule patterns (similar to the robot-files handling of the upcoming section), but this was proved moot by the next update. UPDATE, Jul-2018: server breakage #2 README.txt files have once again vanished from auto-index pages on the server hosting this script due to an unknown GoDaddy Apache-configuration change, and eliminating README.txt files in the rewrite pattern had no effect. Hence, _README.txt files and their fix-readmes.py script have been reinstated (in .htaccess files). Lesson: Apache servers are brittle, and hosting providers are worse. UPDATE, Apr-2020: a new server home Due to poor response time at its former GoDaddy host, this script's site has finally moved to an AWS Lightsail VPS. Among other things, the new host means that root-level server configuration can subsume .htaccess files, and the README issues noted here are now largely legacy. If you wish to adapt this script, your hosting mileage may naturally vary. This new server also runs Python 3.X, which forced multiple patches described ahead. ------------------------------------------------------------------------------- HANDLING CRAWLER ROBOTS.TXT FILES ------------------------------------------------------------------------------- If your site uses a "robots.txt" file to give guidance to crawlers, you MAY want to configure your Apache rewrite rules to avoid routing them to this script for formatting. Otherwise, crawlers may invoke a URL like this and receive an HTML page in response: https://learning-python.com/cgi/showcode.py?name=robots.txt To avoid this, either expand the match pattern to disqualify this filename, or add a rule to match the name and prune further rewrite processing if possible (e.g., with an L or END action code, where they apply). For example, the following rule, which is the form that this site's main .htaccess file actually uses, successfully excludes robot files by using a lookahead negative assertion with a nested non-capturing alternation, plus two capturing groups (yes, yuck): rewriterule ^(?!(?:cgi\/|.*robots.txt))(.*)\.(py|txt|pyw|sh|c|h)$ "https\:\/\/learning-python\.com\/cgi\/showcode\.py\?name\=$1.$2" [NC] The following alternative, though, placed before the showcode rewrite rule did not work on the site server, for reasons TBD (this is subtle business): rewriterule ^(.*)robots.txt$ "https\:\/\/learning-python\.com\/$1robots.txt" [L] If your site does NOT use a robots.txt (and this script's site does not), you probably don't need to care: the error-reply HTML page this script issues when a missing robots.txt is requested should be harmless to your search visibility and crawling results. Per: https://developers.google.com/search/reference/robots_txt#file-format unrecognized content in HTML replies is simply ignored; which makes the reply equivalent to an empty file; which is the same as no robots.txt at all; which means "crawl everything here." Redirects may also be followed, if your showcode rewrite rule uses one (this site's doesn't). Note that it's possible that crawlers may still recognize the directives text in a robots file even if it HAS been formatted as HTML for display by this script. This would make the above tricks unnecessary, but was not tested because this site doesn't use these files; your site may vary. Disclaimer: this is based on Google behavior which other crawlers may or may not mimic, and your robots.txt resolution may have to be applied to any other admin files on your site that match the showcode rewrite rule (e.g., sitemaps?). While this script could support a list of such files forcibly returned as plain-text or 404 error codes (see sitesearch.py), it's easier to delegate to servers by coding rules to exclude such files. ------------------------------------------------------------------------------- UNICODE POLICIES HERE ------------------------------------------------------------------------------- When loading code from files, this script tries a set of Unicode encodings in turn, until one works or all fail. Most Python and text files on this site are UTF-8 (or its ASCII subset), but a few cp1252 and Latin-1 files crop up as examples. The UNICODE_IN encodings list reflects this, but may be changed for use elsewhere (see also the next section). Once loaded, code text is just decoded code points in memory, and is always output as UTF-8-encoded bytes in reply pages. Doing so portably for both Python 3.X and 2.X is possible but subtle; see the Jun-26-18 notes ahead. The next section goes into more detail on one consequence of the Unicode policies applied to displayed file content, and the section following it explores additional Unicode concerns addressed for filenames. ------------------------------------------------------------------------------- AVOID MIXED UNICODE CONTENT ENCODINGS [Sep-2018] ------------------------------------------------------------------------------- [There is a polished and expanded version of the following note online at https://learning-python.com/post-release-updates.html#showcodeunicode.] When using this script, the content of a site's displayable text files should generally all use a common Unicode encoding type (e.g., UTF-8) for reliable display. Else, it's possible that some files may be loaded per an incorrect encoding if their data passes under other schemes. This is especially possible if files use several incompatible 8-bit encoding schemes: the first on the encodings list that successfully loads the data will win, and may munge some characters in the process. This issue cropped up in an older file created with the CP-1252 (a.k.a. Windows-1252) encoding on Windows, whose tools have a nasty habit of silently using its native encodings. This file's slanted quotes failed to display correctly in showcode because Python happily loads the file as Latin-1 (a.k.a. ISO-8859-1), despite its non-Latin-1 quotes. The loaded text encodes as UTF-8 for transmission, but decodes with junk bytes. Here's the story in code. Python does not allow the character '“' to be _encoded_ as Latin-1, in either manual method calls or implicit file-object writes. This quote's 0x201c Unicode code point maps to and from byte value 0x93 in Windows' CP-1252, but is not defined in Latin-1's character set: >>> c = '“' # run in Python 3.X >>> hex(ord(c)) # same in 2.X (using u'“', codecs.open(), print) '0x201c' >>> c.encode('cp1252') # valid in CP-1252, but not Latin-1 b'\x93' >>> c.encode('latin1') UnicodeEncodeError: 'latin-1' codec can't encode character '\u201c'... Conversely, _decoding_ this character's CP-1252 byte to Latin-1 works both in manual method calls and file-object reads. This is presumably because byte value 0x93 maps to an obscure and unprintable "STS" C1 control character in some Latin-1 definitions, though the decoder may simply allow any 8-bit value to pass. It's not a CP-1252 quote in any event: >>> b = b'\x93' >>> b.decode('cp1252') # the proper translation '“' >>> b.decode('latin1') # but it's not a quote in latin1 '\x93' >>> n = open('temp', 'wb').write(b) >>> open('temp', encoding='cp1252').read() '“' >>> open('temp', encoding='latin1').read() # <= what showcode did '\x93' This is problematic in showcode, because this script relies on encoding failures to find one which matches the data and translates its content to code points correctly. Because a CP-1252 file loads without error as Latin-1, its UTF-8 encoding for reply transmission is erroneous; the quote's code point never makes the cut: >>> b.decode('cp1252').encode('utf8').decode('utf8') # load, reply, browser '“' >>> b.decode('latin1').encode('utf8').decode('utf8') # the Latin-1 munge... '\x93' The net effect turns the quote into a garbage byte that browsers simply ignore (it's a box in Firefox's view-source, but is otherwise hidden). If your non-UTF-8 files are _only_ CP-1252, replacing Latin-1 with CP-1252 in the encodings list fixes the issue. However, if your site's files use multiple encodings whose byte ranges overlap but map to different characters, using CP-1252 may fix some files but break others. Latin-1 files using the 0x93 control code, for example, would sprout quotes when displayed (unlikely, but true). The real issue here is that content of mixed encodings is inherently ambiguous in the Unicode model. The _better solution_ is to make sure your site's displayable text files don't use incompatible encoding schemes. At showcode's site, the simplest fix was to adopt UTF-8 as the site-wide encoding, by opening its handful of CP-1252 files as CP-1252, and saving as UTF-8. The set of suspect files can be easily isolated by trying UTF-8 opens (e.g., in an os.walk() loop). Converting to UTF-8 universally will not only help avoid corrupted text in showcode, it might also avoid issues in text editors that are given or guess encoding types. If you give the wrong encoding to an editor, saves may corrupt your data. If you expect a tool to deal with mixed encoding types, guessing may be its only recourse. But guessing is overkill; is impossible to do accurately anyhow; and is not science. Skip the drama and convert your files. UPDATE: preset to 'cp1252' In light of all the above, 'latin1' was eventually replaced by 'cp1252' in showcode's preset input-encodings list, to accommodate a few files at this site that are intentionally not UTF-8 (this is similar in spirit to the policies for parsing web pages in HTML5). CP-1252 is a superset of Latin-1 and should work more broadly, but change as needed for your site's files. This is still only a partial solution for mixed-content ambiguity; use a common Unicode type to avoid encoding mismatches altogether. FOOTNOTE: Latin-1 pass-through Subtly, some scripts, including this site's genhtml web page builder (see learning-python.com/genhtml.html), can often get away with treating CP-1252 and other 8-bit encodings as Latin-1, because bytes whose interpretations differ between the two are passed through unchanged from load to save. What Latin-1 reads and writes as 0x93 is still '“' to CP-1252, though the equivalence falls apart when comparing non-Latin-1 text: >>> '“'.encode('cp1252').decode('latin1').encode('latin1').decode('cp1252') '“' >>> '“'.encode('cp1252').decode('latin1') == '“' # cp1252's meaning lost False This doesn't help in showcode, because data loaded as Latin-1 is not written again as Latin-1; encoding as UTF-8 in the reply makes the munging permanent. FOOTNOTE: bytes processing In some use cases, it's also possible to sidestep Unicode encoding dilemmas altogether by processing files in bytes (not text) mode. This works if the use case does not need to support text matches for arbitrary Unicode keys (genhtml does), and does not need to inform a browser of encodings to properly display text (this script does). Replacing an all-ASCII string of bytes in mixed-encoding files, for example, can get by with bytes mode as long as all the files store ASCII text as ASCII bytes (but UTF-16 won't!); see ip-anon.py. FOOTNOTE: encoding guesses But if you really must guess Unicode encodings of text content in a Python program, you can try with the "chardet" third-party library. Read all about it at https://chardet.readthedocs.io/en/latest/usage.html, and fetch it at https://pypi.org/project/chardet/. This may be useful in some contexts, but its results still are a guess, and come with confidence factors. How in the world did the text story in computing become this convoluted?... ------------------------------------------------------------------------------- MORE UNICODE CONSIDERATIONS HERE [Sep-2020] ------------------------------------------------------------------------------- Beyond reply-body content (the displayed code/text), the following three items describe issues encountered with non-ASCII filenames in both system calls and stdout prints in Python 3.X. These default to ASCII in CGI scripts only, and require special handling here. NON-ASCII FILENAMES, FIX #1: os.stat As of August 2020, this script better supports non-ASCII Unicode characters in filenames on Linux, by manually encoding the file's pathname to bytes per UTF-8, prior to passing it to file system calls in Python 3.X (only). Before this new manual encoding, a to-ASCII encoding run within os.stat() (called from os.path.isfile()) triggered UnicodeEncodeError exceptions for character codepoints outside the ACSII range; the net result produced server-error reply pages for any non-ASCII filename requested. This fix presumably works because Python's default on the Linux server is ASCII, and the patch prevents erroneous to-ASCII auto encoding in os.stat() (but oddly, os.stat() exceptions did not occur in 3.X when running either equivalent code or this same script with MOCK_SERVER=1 directly from a Linux shell on the same machine; see CAUSE below for a later explanation of this). The new manual encode is neither required nor used in Python 2.X, because paths are already encoded bytes (str), not decoded codepoints (Unicode); this site's recent move from a 2.X to 3.X host probably spawned the failures. Caveat: this fix should be harmless outside Linux and Python 3.5, and has been verified offline on both Python 2.X and 3.X, but broader verification remains a to-do. Search on "Aug-20" and "aug20" for code changes. NON-ASCII FILENAMES, FIX #2: header print As of September 2020, this script now properly formats non-ASCII filenames in HTTP reply headers per web standards. Without this, the reply code's print(contenthdr) would raise exceptions for non-ASCII filenames and yield 5xx server replies. This happened only for "view" and "save" requests, which return a filename in headers. This also was observed only for Python 3.X on Linux (and again seems to reflect Linux defaults; see CAUSE below), but the new formatting both avoids print() errors and invokes correct handling in clients that receive the headers. Search for "Sep-2020" here for more on the changes. NON-ASCII FILENAMES, CAUSE: CGI locales After further research, it's now known that the 3.X non-ASCII filename issues fixed in this script stem from the fact that the locale environment settings used at the terminal are not available in the CGI context, which causes encodings to default to ASCII in CGI scripts *only*. Hence, CGI code that works well in testing can fail live, and code that is never run as a CGI script won't have the same problems (yes, yuck). The manual-encoding work-arounds here solve the failures tactically, though strategic approaches (e.g., PYTHONIOENCODING settings in .htaccess) may help too. For more on this topic, search the web for "python cgi stdout unicode encoding" or see: https://stackoverflow.com/questions/9322410/set-encoding-in-python-3-cgi-scripts. ------------------------------------------------------------------------------- DON'T DISPLAY FILES OUTSIDE THE WEBSITE FOLDER [Feb-2022] ------------------------------------------------------------------------------- As a security measure, this script no longer shows files located outside the website's root folder, unless configured to do so. To enable this check, set the variable SITE_ROOT_FOLDER ahead to the path to your site's folder. To disable this check, set this to a broader path (e.g., '/'). When enabled, all requested files will be checked to ensure that they are nested in the site folder, by running both your folder setting and the requested path through Python's os.path.realpath (which is like abspath, but evaluates any symlinks). Any requested item whose path is not nested in the site folder's path will be denied, with an error-message page. This is done to deny dubious requests to view files outside the site. A recent analytics hit revealed that someone requested '../../../../etc/passwd' by passing it as a parameter in an explicit URL. The prior version of this script happily returned the host system's passwd file in response, though it was fully useless to the requester (as usual, all real passwords are in the shadow file, which cannot be accessed, and isn't plain text in any event). This passwd file worked only because it is world readable; other system files are naturally off limits. This script also fetches only files relative to its own folder, and there is no way to view folder listings with this script, so the requester had to know the structure of a bitnami install; creepy, but true. This new version of this script avoids such security risks for sensitive but accessible files, and frustrates similar prying eyes, by disallowing all requests for anything outside the site's own folder with the setting ahead. It relies on realpath as is, and could probably check for inodes in paths too or instead (see stat), but this seems unwarranted here; tbd. ------------------------------------------------------------------------------- OTHER USAGE IDEAS ------------------------------------------------------------------------------- This script can also be invoked by URL in the "action" tag of a form in an HTML page; could be submitted by a script (see Python's urllib); and might work as an Apache handler (to be explored). ------------------------------------------------------------------------------- CAVEATS SUMMARY ------------------------------------------------------------------------------- As is, this script reflects a number of tradeoffs: -Its code must run on Python 2.X and 3.X, as hosting servers vary widely. -Its footer code must avoid copies of text normally generated by genhtml. -Its error checking is minimal, as it is used only in well-known contexts. -Its ".." assumption for subject files' paths is not very general. -Its Apache rewrite rule breaks "README.txt" in index pages (see above). -Its Apache rewrite rule may complicate robots.txt handling (see above). -Its always-UTF-8 output policy means others are converted to this on saves. -Its Unicode encodings list may fail in mixed-type contexts (see above). -Its manual UTF-8 encoding of filenames may have consequences (see above). OTOH, it works as intended, and demos CGI; expand and improve as desired. =============================================================================== """ #################### # CODE STARTS HERE # #################### # 2.X: not needed for print('onestr'), but is for trace=print from __future__ import print_function import cgitb; cgitb.enable() # route python exceptions to browser/client import cgi, os, sys, codecs UsingPython3 = sys.version[0] == '3' UsingPython2 = sys.version[0] == '2' if UsingPython3: # py 3.X/2.X compatible from html import escape as html_escape # run on 2.X only initially from urllib.parse import quote_plus # moved to 3.X server later elif UsingPython2: from cgi import escape as html_escape # for text added to HTML from urllib import quote_plus # for text added to URL else: assert False, 'The future remains to be written' #============================================================================== # Switches and constants #============================================================================== #------------------------------------------------------------------------------ # Jun-26-18 note: Browsers allow "UTF8" but "UTF-8" is technically the HTML # encoding name; python allows synonyms, including both "UTF-8" and "utf8". # For background, see: https://encoding.spec.whatwg.org/#names-and-labels; # # Sep-01-18 note: See AVOID MIXED UNICODE ENCODINGS above - 'latin1' was # replaced with 'cp1252' in UNICODE_IN for the host site as a half-measure. # # See [Feb-2022] above: set SITE_ROOT_FOLDER to your site's root-folder path # to prevent access to items outside the folder; set to '/' to allow them. #------------------------------------------------------------------------------ MOCK_VALUES = False # 1=simulate parsed inputs MOCK_SERVER = False # 1=simulate client request UNICODE_IN = ['UTF8', 'cp1252'] # try in turn for code file content UNICODE_OUT = 'UTF-8' # for text in generated reply page TEMPLATE = 'showcode-template.txt' # the reply-page format FOOTER = '../dummy-footer.html' # site-wide footer code SITE_ROOT_FOLDER = '/opt/bitnami/apache2/htdocs' # made canonical via realpath trace = lambda *args: None # or print, to display on stdout #============================================================================== # Get input filename (and raw-text mode?) sent from the client #============================================================================== # Parse and/or forge request if MOCK_VALUES: # simulate parsed request for testing class Mock: def __init__(self, value): self.value = value form = dict(name=Mock('timeformat.py')) # + rawmode=Mock('view')? else: if MOCK_SERVER: # jun18: simulate post-server, pre-pylibs state os.environ['CONTENT_TYPE'] = 'Content-type: application/x-www-form-urlencoded' os.environ['QUERY_STRING'] = ( 'name=pyedit-products/unzipped/docetc/examples' #'/aaa"bbb"ccc' '/Non-BMP-Emojis/Non-BMP-Emoji-both-%f0%9f%98%8a.txt' #'&rawmode=view' ) form = cgi.FieldStorage() # parse form/url input data # Extract request inputs if 'name' not in form: name = 'cgi/showcode.py' # show myself: more useful """ # error check: custom reply = hdr + blankline=\n + msg print('Content-type: text/plain\n') print('Please provide a value for "name" in the request.') sys.exit(1) """ else: name = form['name'].value # real or mocked, pathname relative to '..' if 'rawmode' not in form: rawmode = False else: rawmode = form['rawmode'].value # 'view' or 'save' or absent=formatted #============================================================================== # Load the code from a file in "..", in 1 of N Unicode encodings #============================================================================== #------------------------------------------------------------------------------ # "name" may be a basename or a pathname relative to ".." (site root). # Both open()/read() flavors retain \r on Windows, decode to code points, # and return a Unicode object: a py2 u'xx' unicode, or a py3 'xx' str. # Tries N Unicode types for input, but always outputs as UTF-8 bytes. # os.path.isfile() is a superset of os.path.exists(): don't need both. # File loads may fail due to Unicode errors or permissions (e.g., 0200); # # Aug-20: In Python 3.5, the os.stat() call in os.path.isfile() raised a # UnicodeEncodeError encoding exception for filenames having non-ASCII # Unicode characters (e.g., both 16-bit BMP symbols, and larger emojis). # To fix, now manually encodes to UTF-8 bytes first, which prevents the # auto ASCII encoding that failed. The encode isn't required and even # fails in Python 2.X (this site's former server version), because path # is already encoded bytes (2.X str), not decoded codepoints (2.X unicode). # See the top-of-file docstring's "UNICODE POLICIES HERE" for more details, # especially its "NON-ASCII FILENAMES, CAUSE" for more on the exception; # in short, CGI scripts lack locale settings, so encodings default to ASCII. # # Feb-22: see [Feb-2022] docs at top of file - now denies requests for # all files outside the website folder, unless it's set to '/' or similar. # The new code also has to take care to avoid mixing string types in 3.X. #------------------------------------------------------------------------------ path = '..' + os.sep + name # _decoded_ str in 3.X, _encoded_ str in 2.X # aug20: prevent to-ASCII auto encoding in 3.X's os.stat() on Linux if UsingPython3: path = path.encode('utf8') # convert str codepoints to preencoded bytes # feb22: avoid mixed-type errors for ops ahead if UsingPython3: SITE_ROOT_FOLDER = SITE_ROOT_FOLDER.encode('utf8') # feb22: deny access to files outsite site folder realsite = os.path.realpath(SITE_ROOT_FOLDER) realpath = os.path.realpath(path) if not realpath.startswith(realsite): code = (u'Error: requested file is outside the website folder.\n\n' 'Please request only files that are part of this site.\n') # feb22 elif not os.path.isfile(path): code = (u'Error: file does not exist or is not a file.\n\n' 'Please verify the filename in your request.\n') # apr18 else: for tryenc in UNICODE_IN: try: if UsingPython3: code = open(path, mode='r', encoding=tryenc, newline='').read() else: code = codecs.open(path, mode='r', encoding=tryenc).read() except: pass # try next encoding on list else: break # load successful: skip else else: # none worked code = (u'Error: could not open file.\n\n' "Please adjust the script's UNICODE_IN list " # jun18 "or the file's permissions.\n") # jun19 # Sep-2024: drop one empty line at end of code for displays if code.endswith('\n\n'): code = code[:-1] # all but last \n (or rstrip?) #============================================================================== # Load and expand the HTML template #============================================================================== #------------------------------------------------------------------------------ # It's okay to load the template as str, even though "code" is unicode: # in py3 they're the same - both are str, which is always Unicode text; # in py2 they differ, but str is coerced up - '%s' % u'spam' => u'spam', # even for dicts: see learning-python.com/showcode-unicode-demo.txt. # # Jun-26-18 - BUT we must decode strs to Unicode for py2 if they may be # non-ASCII, because py2 expects strs to be all-ASCII whenever mixed with # unicodes. This applies only to "namehtml", a str in 2.X encoded as UTF-8 # per web conventions: "code" is already unicode, "nameurl" URL-escapes any # non-ASCIIs, and "footer" and "template" are assumed to be ASCII files. # Even if name is decoded, error-message "code" must be all-ASCII str or # unicode too - use u'xxx' above in both 2.X and 3.X (3.X requires 3.3+). #------------------------------------------------------------------------------ if rawmode: reply = code # send text as is (after encoding to bytes) else: template = open(TEMPLATE).read() # template file in '.', ASCII only codehtml = html_escape(code) # HTML-escape any characters special in HTML namehtml = html_escape(name) # template also hardcodes some URL escapes # no longer need to strip 'cgi/' here: used in URL query, not raw link nameurl = quote_plus(name) # URL-escape this: added to query in template # site-specific footer handling footer = open(FOOTER).read() # load dummy ASCII generated footer html in .. for link in ('HREF', 'href', 'SRC'): # munge it to add ".." to all nested item refs old = '%s="' % link # this avoids copying code (see template text) new = '%s="../' % link footer = footer.replace(old, new) for undo in ('mailto', '#'): # undo up-rerouting for two special-cases new = 'HREF="../%s' % undo # still beats maintaining copied code... old = 'HREF="%s' % undo footer = footer.replace(new, old) trace('namehtml:', type(namehtml), repr(namehtml)) trace('codehtml:', type(codehtml)) # jun-26-18: allow non-ASCII filenames in Python 2.X (see above) if UsingPython2: namehtml = namehtml.decode(UNICODE_OUT) trace('namehtml2:', type(namehtml), repr(namehtml)) reply = template % dict( # unicode reply: replace template targets __NAME__ = namehtml, # the sent and escaped filename __NAMEURL__ = nameurl, # the filename for raw-text link __CODE__ = codehtml, # the loaded and escaped Unicode text __FOOTER__ = footer) # the munged dummy generated toolbar html trace('reply:', type(reply), repr(reply[:40])) #============================================================================== # Print the reply stream back to the client #============================================================================== #------------------------------------------------------------------------------ # Write UTF8-encoded bytes, use "charset" to force Unicode type to match. # "inline" is always view, but may require cmd/ctl-A+C to save contents. # "attachment" is usually save, but opens may fail on some platforms, and # this is just view on others (notably, iOS: there's no user file access). # # Mar-2020: Usage on an AWS Lightsail VPS uncovered a bug under Python 3.X: # stdout has to be flushed after print()s here, else headers are output # last - after replybytes. This is due to buffering, and can be recreated # by "python3 showcode.py | more" to simulate Apache's flavor. It was found # on an Ubuntu Bitnami LAMP stack running Python 3.5.2; it went unseen on a # former host using Python 2.X, as well as on Mac OS with PYTHONUNBUFFERED=1. # # Sep-2020: Non-ASCII filenames in HTTP reply parameters created for Raw text # "view" and "save" requests made print() raise exceptions in Python 3.X on # Linux (and yield 5xx server replies). To fix, instead of using a simple # format [filename="%s"], encode non-ASCII filenames per web standards here: # https://tools.ietf.org/html/rfc5987 (original) # https://tools.ietf.org/html/rfc8187 (updated) # https://tools.ietf.org/html/rfc2616#section-2.2 # This both ensures that filename parameter values are ASCII so that print() # works in 3.X/Linux, and invokes proper handling in recipient clients. # See "NON-ASCII FILENAMES, CAUSE" above for more on why this is required; # in short, CGI scripts lack locale settings, so stdout defaults to ASCII. #------------------------------------------------------------------------------ # Sep-2020 if hasattr(str, 'isascii'): isascii = str.isascii # py 3.7+ only: use builtin method else: def isascii(text): try: text.encode('ascii') # all others: use custom function except: return False else: return True def httpHeaderFormat(text): """ ------------------------------------------------------------------------------ Sep-2020: Format non-ASCII parameters in HTTP replies to ASCII per standard. Note that for non-ASCII this formats encoded bytes, not decoded codepoints. ASCII is formatted to "..." as before, but any embedded " are now \" escaped. Non-ASCII example: ... filename*=UTF-8''%c2%a3%20spam%20%e2%82%ac%20eggs ... ASCII example: ... filename="spam\"and\"eggs" ... ------------------------------------------------------------------------------ """ trace('text:', type(text), repr(text)) # 3.X: decoded str, 2.X: encoded str if isascii(text): return '="%s"' % text.replace('"', '\\"') else: if UsingPython2: text_encoded = text # it's already encoded bytes else: text_encoded = text.encode('utf8') # Unicode codepoints => bytes text_escaped = quote_plus(text_encoded) # percent escapes return "*=UTF-8" + "''" + text_escaped # HTTP extended notation (no lang) if not rawmode: contenthdr = 'Content-type: text/html; charset=%s' % UNICODE_OUT else: dispostype = 'inline' if rawmode == 'view' else 'attachment' basename = os.path.basename(name) baseparam = httpHeaderFormat(basename) # sep20 contenthdr = 'Content-type: text/plain; charset=%s\n' % UNICODE_OUT contenthdr += 'Content-Disposition: %s; filename%s' % (dispostype, baseparam) replybytes = reply.encode(UNICODE_OUT) # send encoded bytes: print is iffy print(contenthdr) # reply = hdrs + blankline + html print('') # need '' for 2.X if no __future__ if UsingPython2: sys.stdout.write(replybytes) # py2 accepts a str for the bytes else: sys.stdout.flush() # mar20: else headers last on 3.X! sys.stdout.buffer.write(replybytes) # py3 stdout is str: use io layer # And now it's up to the client