File: genhtml/genhtml.py
#!/usr/bin/python3 """ =========================================================================================== genhtml.py - static HTML inserts Version: 2.7, March 2022 (see VERSIONS ahead) License: provided freely but with no warranties of any kind Attribution: © M. Lutz (learning-python.com), 2015-2022 ------------------------------------------------------------------------------------------ SYNOPSIS ------------------------------------------------------------------------------------------ genhtml builds HTML files by performing key-based text insertions on all the files in a folder (dir) at site-build time. Given an HTML templates dir and an HTML inserts dir, it generates final HTML files by applying text replacements to the templates, where replacement keys and text correspond to insert file names and contents. The text which replaces keys in the generated files can be any textual component: HTML, CSS, JavaScript, and so on. For static insert content (common text that can change over time, but need not be generated on each new page request), the net result provides a basic webpage macro utility, and is an alternative to: - Mass file edits on every common-item change - which can be painfully tedious - Client-side includes via embedded JavaScript - which not all visitors may run - Server-side includes via PHP or Apache - which require a server to view pages - cpp - which may not be present, plus makefiles - which must be manually coded This script adds a local admin step, as it must be run on every HTML content change (e.g., from a publishing script), but there is no direct HTML include. ------------------------------------------------------------------------------------------ COMMAND-LINE USAGE ------------------------------------------------------------------------------------------ Given the SOURCEDIR, TARGETDIR, and INSERTDIR settings in this script's code: % python3 genhtml.py (or run via icon or shortcut click) 1) Copy all changed non-HTML files in SOURCEDIR to TARGETDIR (if any) 2) Regenerate all HTML files whose SOURCEDIR template or any INSERTDIR insert file it uses has changed since the HTML's latest generation, replacing all insert-file references with insert-file text, and storing the expanded HTML files in TARGETDIR % python3 genhtml.py [filename]+ Same, but apply to just one or more SOURCEDIR files, listed without dir name ------------------------------------------------------------------------------------------ USAGE PATTERNS ------------------------------------------------------------------------------------------ As shipped, SOURCEDIR, INSERTDIR, and TARGETDIR all are in the script's current working directory (a.k.a. '.', from where the script is run); change as desired. To use this script to maintain a site's files: 1) Change HTML template files in SOURCEDIR, and/or insert files in INSERTSDIR. 2) Run this script to regenerate all changed HTML files in TARGETDIR as needed. 3) Upload newly generated files (or all) from the TARGETDIR to the web server. Do not change HTML files in TARGETDIR: they may be overwritten by generations! There are two ways to structure a site's files: A) Keep both HTML templates and all other site files in SOURCEDIR. In this mode, changed non-HTML files are copied to TARGETDIR when HTML files are regenerated. B) Use SOURCEDIR for HTML template files only, keep other web site files in TARGETDIR. This mode avoids copying other non-HTML files to TARGETDIR on HTML regenerations. Either way, TARGETDIR is always the complete web site, for viewing and uploads. ------------------------------------------------------------------------------------------ TEXT REPLACEMENTS ------------------------------------------------------------------------------------------ Text replacements are flexible, and derived from insert folder content: - Replacement keys are '$XXX$' for all INSERTDIR/XXX.txt filenames. - Replacement values are the contents of the INSERTDIR/XXX.txt files. - Algorithm: For each changed '*.htm' and '*.html' (caseless) HTML template in SOURCEDIR: For each XXX in INSERTDIR/XXX.txt: Replace any and all '$XXX$' in HTML template with the content of file INSERTDIR/XXX.txt Save the result in TARGETDIR Other changed non-HTML files (if any) are copied to TARGETDIR verbatim. To automate changing DATES in both HTML files and insert files, the script also replaces special non-file '$_DATE*$' keys: e.g., '$_DATELONG$' => 'November 6, 2015'. See the script's code ahead for the full set of date keys available. Example key=>file replacements (with possible nested inserts, described ahead): Coded in <HEAD> $STYLE$ => INSERTDIR/STYLE.txt (a <style> or <link rel...>) $SCRIPT$ => INSERTDIR/SCRIPT.txt (analytics or other JS code block) $ICON$ => INSERTDIR/ICON.txt (site-specific icon link spec) Coded in <BODY> $FOOTER$ => INSERTDIR/FOOTER.txt (a standard sites links toolbar) $HEADER$ => INSERTDIR/HEADER.txt (a standard header block) $TOTOC$ => INSERTDIR/TOTOC.txt (a standard go-to-index button line) See also "__docs__/template-pattern.html: for a skeleton use case example file, and the "Html-templates" test folder for additional template examples. ------------------------------------------------------------------------------------------ NESTED INSERTS ------------------------------------------------------------------------------------------ To allow insert files to be built up from other insert files (in an intentionally limited fashion), the script also replaces any '$XXX$' keys in the loaded text of insert files, before regenerating any HTML files. For an example use case, see FOOTER-COMMON.txt and its clients in the Html-inserts folder; it is inserted into other footer insert files which have varying footer parts. For dependency checking, any newer modtimes of nested inserts are also propagated to their inserter (see ahead). Limitation: by design, nesting is only 1-level deep - an HTML template may insert an insert file which inserts other insert files, but no more (this is not recursive). ------------------------------------------------------------------------------------------ AUTOMATIC DEPENDENCY-CHANGE DETECTION ------------------------------------------------------------------------------------------ This script acts like a makefile: changing files suffices to trigger regeneration on the next run. In more detail, an HTML file is automatically generated if its expansion does not yet exist, and regenerated if: (a) Its HTML template file has been changed since the HTML's last generation; or (b) Its HTML template file inserts any file that has been changed since the HTML's last generation; or (c) Its HTML template file inserts any file that inserts any other file which has been changed since the HTML's last generation (nested inserts). In other words, the script generates expanded HTML files for all HTML templates that have no expansion yet; are newer than their expanded versions; or use any insert file that is newer than their expanded versions. Conversely, an HTML file is not regenerated if neither the HTML template nor any used insert is newer than the expansion target. Additional dependency notes: FORCEREGEN To force a regeneration of all HTML files, open and save any insert file used by every template (if any), or set the FORCEREGEN variable below. When True, this switch effectively ignores dependency tests and generates all. DATE INSERTS By design, there is no dependency checking for non-file '$_DATE*$' inserts (else each date-key client would be regenerated every day!). Open and save date-client files to force their regen with updated dates when appropriate. NESTED INSERTS When nested inserts are used, dependencies are transitive: the modtime of any insert file is considered to be the greater of the modtime of the insert file itself, or the modtimes of any nested insert files which the insert file uses. Modtimes are propagated to nested inserters before testing template dependencies. ------------------------------------------------------------------------------------------ CRUFT HANDLING (Jun2017) ------------------------------------------------------------------------------------------ genhtml now handles ".*" macOS/Unix cruft files lurking in inserts and templates folders. This includes both ".DS_Store" Finder files, plus any "._*.{txt, html}" AppleDouble resource-fork files on non-macOS filesystem drives. These files are skipped in inserts so they don't cause errors, and are simply treated like other non-HTML in templates and copied over blindly because more may crop up after this script runs and before the site is uploaded (.DS_Store happens!). To handle cruft files post site generation, filter them out with your zip or upload tools; ziptools (learning-python.com/ziptools.html), for example, skips cruft with its "-skipcruft", and macOS Finder's Compress isolates proprietary items in folders. ------------------------------------------------------------------------------------------ TEMPLATE UNICODE ENCODINGS (Sep2018) ------------------------------------------------------------------------------------------ When loading page templates, this script tries Unicode encodings on a changeable list, and uses the first that succeeds. This is flexible, but may require configuration for your site's files. As an example, this script can generally get away with treating CP-1252 files as Latin-1, because bytes whose decoding 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): >>> c = '“' >>> c.encode('latin1') UnicodeEncodeError: 'latin-1' codec can't encode character '\u201c'... >>> >>> n = open('temp', 'w', encoding='cp1252').write(c) >>> open('temp', 'r', encoding='cp1252').read() '“' >>> L = open('temp', 'r', encoding='latin1').read() # genhtml load >>> L '\x93' >>> n = open('temp', 'w', encoding='latin1').write(L) # genhtml save >>> open('temp', 'r', encoding='cp1252').read() # quote retained '“' >>> open('temp', 'rb').read() b'\x93' More generally, because Latin-1's encoded bytes are also code-point values, 'latin1' often works as well as 'cp1252' and other 8-bit encodings in this program's TemplateEncodings Unicode encodings list, as long as you don't need to match replacement keys containing text outside Latin-1's character set: >>> '“'.encode('cp1252').decode('latin1').encode('latin1').decode('cp1252') '“' >>> '“'.encode('cp1252').decode('latin1') == '“' # cp1252's meaning lost False As guidelines, though, use CP-1252 if you know your files are this type, and convert files to a common site-wide file encoding like UTF-8 to avoid encoding mismatch issues altogether. For more tips on choosing encodings, see the following - an online article with an expanded version of this note in its "Footnote," and showcode.py which uses a similar Unicode-choices scheme: learning-python.com/post-release-updates.html#showcodeunicode learning-python.com/cgi/showcode.py?name=cgi%2Fshowcode.py Both mention the third-party "chardet" as an alternative for encoding guesses. ------------------------------------------------------------------------------------------ MIXED UNICODE ENCODINGS IN TEMPLATES AND INSERTS (Dec2018) ------------------------------------------------------------------------------------------ Subtlety: as coded, genhtml assumes that the content of insert files is compatible with the Unicode encoding used to load and later save template files which use the inserts. This can be an issue if the template file's encoding is too narrow for a used insert's text. For example, if a template file is loaded as ASCII and uses a UTF-8 insert file containing non-ASCII text, the expanded result's save will fail when trying to save as ASCII again. To avoid this potential, either make sure your inserts are Unicode-compatible with template files that use them (e.g., all ASCII, Latin-1, or UTF-8), or change the "TemplateEncodings" list below to accommodate all possible cases. As an example of the latter approach, if this list is set to begin with UTF-8 (not ASCII), it will handle simple ASCII templates and inserts (as a subset encoding), but will also allow for non-ASCII inserts in ASCII templates and save the results as UTF-8. This works because ASCII templates will be loaded as UTF-8, though you may wish to update such files' charset declarations for the broader UTF-8 encoding also used to save the expanded result (this is why saves are still allowed to fail if the load encoding does not work). [2.7] UPDATE MAR-2022: saves are now retried as UTF-8 if template load encodings fail. This allows inserts to differ, but be sure to set the page's encoding too. It cropped up for a non-ASCII insert file and an ASCII template, which may be common in practice. To see saves retried, look for output messages like this: => index.html: ** ascii failed - retrying save as UTF-8: GENERATED, using UTF-8 To disable this and fallback on prior exceptions, set RETRYSAVES to False ahead. ------------------------------------------------------------------------------------------ MISCELLANEOUS NOTES ------------------------------------------------------------------------------------------ - Skips replacement targets not present in HTML text (replace() is a no-op). - Skips replacement targets having no INSERTDIR file (no .replace() is run). - Tries multiple Unicode encodings for HTML text: expand the set as needed. - Assumes insert files are all in same Unicode encoding: change as needed. - Changed external CSS files must be uploaded, but do not require or trigger HTML regeneration here (unlike changed CSS <link>s or inline CSS code inserts). - See "Programming Python, 4th Edition" for automated FTP site upload scripts. - File modtimes are simply floats, giving seconds since the "epoch": >>> import os, time >>> t1 = os.path.getmtime('.') >>> t1 1446572919.58063 >>> time.ctime(t1) 'Tue Nov 3 09:48:39 2015' >>> t1 += 1 >>> time.ctime(t1) 'Tue Nov 3 09:48:40 2015' ------------------------------------------------------------------------------------------ TBDs ------------------------------------------------------------------------------------------ - Nested inserts could be allowed to be arbitrarily deep, rather than limiting them to just one level. This entails substantial change and extra complexity, though, which has not yet been justified by the proprieter's use cases. - Subdirectories are not directly supported, though they can be maintained as separately generated and uploaded working folders (e.g., via Bash scripts). - Automatic "<!-- -->" comment wrappers could be emitted, but they may be invalid for shorter text inserted into other lines (versus text blocks). - It might be useful to parameterize inserts in some fashion. For instance, between '@' delimiters, allow a script name and arguments defining a command line whose stdout gives the insert text. This is an order of magnitude more complex, though, and is not warranted by any use case so far. - Per the Dec2018 note above: should save failures try a broader type like UTF-8; and should the preset TemplateEncodings start with UTF-8 instead of ASCII? Neither is done automatically today, because a load/save encoding difference may require additional user actions (e.g., charset changes). UPDATE - see the Mar2022 update at note above: saves are now retried as UTF-8. ------------------------------------------------------------------------------------------ VERSIONS ------------------------------------------------------------------------------------------ 2.7, Mar-07-22: retry page saves with UTF-8 if template's encoding fails. 2.6, Sep-26-20: docs only - reformatted/rewrote this docstr for readability. 2.6, Dec-07-18: docs only - add note about mixing template/insert encodings. 2.5, Sep-01-18: docs only - add note that Latin-1 handles cp1252 files. 2.5, Feb-28-18: add new date format, use just *.txt inserts (skip dirs, etc.). 2.4, Jun-11-17: skip/handle ".*" macOS cruft files in inserts and templates dirs. 2.3, Jun-03-16: add built-in date format that strips leading "0" from day number. 2.2, Dec-27-15: change summary text to "generated", to match trace displays. 2.1, Dec-26-15: minor patch to propagate modtimes of nested insert files to their nesters correctly when > 1 file is nested (compare to loop max, not nester). 2.0, Nov-26-15: smarter dependency checking, refactor code. Don't regenerate an HTML file for a changed insert file, unless the HTML's template actually USES the insert file, or an insert file that inserts it. Also refactor as functions: at ~250 lines, top-level script code becomes too scattered to read (and at ~1K lines, class structure is nearly required). 1.X, Initially released sometime in 2015 (presumably). =========================================================================================== """ # CODE STARTS HERE import os, sys, shutil, time trace = lambda *args: None # set to print to see more output # user settings (all three dirs are in '.' cwd as shipped) INSERTDIR = 'Html-inserts' # insert text, filename gives key: XXX.txt. SOURCEDIR = 'Html-templates' # load html templates (and others?) from here TARGETDIR = 'Complete' # save expanded html files (and others?) to here CLEANTARGET = False # True = empty all files in TARGETDIR first FORCEREGEN = False # True = regenerate all HTML files, ignoring dependencies RETRYSAVES = True # True = try UTF-8 if template encoding fails on saves [2.7] # customize Unicode encodings iff needed - see Sep2018 and Dec2018 notes above TemplateEncodings = ('ascii', 'utf8', 'latin1', 'utf16') # try each, in turn InsertsEncoding = 'utf8' # use for all inserts def loadInserts(): """ ------------------------------------------------------------------------- Load insert files/keys and modtimes. (Jun2017) Skip any hidden ".*" macOS cruft files lurking in inserts dir, to avoid errors on "._*.txt" AppleDouble or any other Unix hidden files. ------------------------------------------------------------------------- """ inserts, insmodtimes = {}, [] for insertfilename in os.listdir(INSERTDIR): # Added Jun2017 if insertfilename.startswith('.'): # skip macOS (and Unix) cruft continue # Added Feb2018 # skip non-text files, dirs if not insertfilename.endswith('.txt'): continue # use just *.txt inserts insertkey = '$' + insertfilename[:-4] + '$' # key='$XXX$' from 'XXX.txt' try: path = os.path.join(INSERTDIR, insertfilename) # load insert text for key file = open(path, encoding=InsertsEncoding) # platform default or custom text = file.read() file.close() # close for non-CPython inserts[insertkey] = text insertmodtime = os.path.getmtime(path) # modtime for changes test insmodtimes.append([insertkey, insertmodtime]) # add file-based inserts only except: inserts[insertkey] = '' # empty if file error return inserts, insmodtimes def addDateInserts(inserts): """ ------------------------------------------------------------------------- Add special non-file replacement keys (evolve me). Not file-based, so never added to insmodtimes list. ------------------------------------------------------------------------- """ inserts['$_DATELONG$'] = time.strftime('%B %d, %Y') # 'November 06, 2015' inserts['$_DATESHORT$'] = time.strftime('%b-%d-%Y') # 'Nov-06-2015' inserts['$_DATENUM$'] = time.strftime('%m/%d/%Y') # '11/06/2015' inserts['$_DATETIME$'] = time.asctime() # 'Fri Nov 6 10:44:58 2015' inserts['$_DATEYEAR$'] = time.strftime('%Y') # '2015' # Added Jun2016 shortenday = time.strftime('%b-%%s-%Y') shortenday %= time.strftime('%d').lstrip('0') # drop day leading 0 inserts['$_DATESHORTER$'] = shortenday # 'Jun-3-2016' # Added Feb2018 # drop day leading 0 too try: shortenedday = time.strftime('%B %-d, %Y') # 'November 6, 2015' inserts['$_DATELONG2$'] = shortenedday # alt technique: -d except: inserts['$_DATELONG2$'] = inserts['$_DATELONG$'] # if not supported def propagateModtimes(insmodtimes): """ ------------------------------------------------------------------------- For nested inserts, the modtime of any insert file is considered to be the greater (later) of that of the file itself and that of any file it inserts. Propagate newer modtimes from nested inserts to their clients before expanding nested inserts or HTML templates. ------------------------------------------------------------------------- """ for pair1 in insmodtimes: # pairs are mutable lists [key1, modtime1] = pair1 latest = modtime1 for [key2, modtime2] in insmodtimes: if (key2 in inserts[key1]) and (modtime2 > latest): # dec15: not modtime1 latest = modtime2 pair1[1] = latest def expandInserts(inserts): """ ------------------------------------------------------------------------- Globally replace any keys in loaded insert-file text. Expands any nested inserts before inserts applied to html. ------------------------------------------------------------------------- """ for key1 in inserts: # for all insert texts text = inserts[key1] for key2 in inserts: # for all insert keys text = text.replace(key2, inserts[key2]) # no-op if no match inserts[key1] = text # inserts changed in-place def sourceNewer(pathfrom, pathto, allowance=2): """ ------------------------------------------------------------------------- Was pathfrom changed since pathto was generated? 2-seconds granularity needed for FAT32: see Mergeall. This and its follower assume both file paths exist. ------------------------------------------------------------------------- """ fromtime = os.path.getmtime(pathfrom) totime = os.path.getmtime(pathto) return fromtime > (totime + allowance) def insertsNewer(textfrom, pathto, insmodtimes, allowance=2): """ ------------------------------------------------------------------------- Was any used insert file changed since pathto was generated? 2-seconds granularity needed for FAT32: see Mergeall. This could use any() and generators... but should it? ------------------------------------------------------------------------- """ # for all insert files, check if newer and used totime = os.path.getmtime(pathto) for (inskey, instime) in insmodtimes: if instime > (totime + allowance) and inskey in textfrom: return True return False def loadTemplate(pathfrom): """ ------------------------------------------------------------------------- Try to load an HTML template file, using various Unicode types, from simpler to more complex. Return tuple of text + encoding. See also "TEMPLATE UNICODE ENCODINGS" above. ------------------------------------------------------------------------- """ for encoding in TemplateEncodings: try: file = open(pathfrom, mode='r', encoding=encoding) text = file.read() return (text, encoding) # success: return now except: trace(encoding, 'invalid, ', sys.exc_info()[0]) return (None, None) # no encoding worked def generateHtmls(filestoprocess, inserts, insmodtimes): """ ------------------------------------------------------------------------- Generate expanded HTML files for all HTML templates that have no expansion yet, are newer than their expanded versions, or use any insert files which are newer than the templates' expanded versions. (Jun2017) Copy any ".*" macOS/Unix cruft files like non-HTML items, to avoid errors for "._*.html" (filter out when zip or upload). ------------------------------------------------------------------------- """ global numcnv, numcpy, numskip, numfail for filename in filestoprocess: print('=>', filename, end=': ') pathfrom = os.path.join(SOURCEDIR, filename) pathto = os.path.join(TARGETDIR, filename) if not os.path.isfile(pathfrom): # skip any subdirs, etc. print('non-file, skipped') numskip += 1 elif filename.startswith('.') or not filename.lower().endswith(('.htm', '.html')): # # non-html file: don't attempt regen, copy to target if changed; # used when entire site in templates dir, not just templates; # do macOS ".*" cruft here - skip in ziptools/uploadall (Jun2017); # if os.path.exists(pathto) and not sourceNewer(pathfrom, pathto): # source file unchanged, don't copy over print('unchanged, skipped') numskip += 1 else: # copy in binary mode unchanged rawbytes = open(pathfrom, mode='rb').read() file = open(pathto, mode='wb') file.write(rawbytes) file.close() # close for non-CPython shutil.copystat(pathfrom, pathto) # copy modtime over too print('COPIED unchanged') numcpy += 1 else: # # html file: regen to target if html or used inserts changed; # whether templates dir is entire site, or template files only; # macOS "._*.html" cruft files won't reach here (loads won't fail); # (text, encoding) = loadTemplate(pathfrom) if text == None: print('**FAILED**') numfail += 1 elif (os.path.exists(pathto) # target generated and not ( sourceNewer(pathfrom, pathto) or # source not changed insertsNewer(text, pathto, insmodtimes) or # no used insert changed FORCEREGEN # not forcing full regen )): # neither html template nor any used insert newer than target print('unchanged, skipped') numskip += 1 else: # globally replace keys in text and copy over; # no copystat(): insertsNewer() needs new modtime for key in inserts: # for all filename keys text = text.replace(key, inserts[key]) # no-op if no match # 2.7: try utf8 for save if template encoding fails: inserts may differ resaved = False try: file = open(pathto, mode='w', encoding=encoding) # encoding=template's file.write(text) except UnicodeEncodeError: if not RETRYSAVES: raise else: resaved = True print('\n\t** %s failed - retrying save as UTF-8' % encoding, end=': ') file.close() file = open(pathto, mode='w', encoding='utf8') file.write(text) file.close() # close for non-CPython print('GENERATED, using', encoding if not resaved else 'UTF-8') numcnv += 1 if __name__ == '__main__': """ ------------------------------------------------------------------------- Top-level code run on script invocation. ------------------------------------------------------------------------- """ numcnv = numcpy = numskip = numfail = 0 # globals all # empty target dir? if CLEANTARGET: for filename in os.listdir(TARGETDIR): os.remove(os.path.join(TARGETDIR, filename)) print('--Target dir cleaned') # load/add inserts text inserts, insmodtimes = loadInserts() addDateInserts(inserts) print('Will replace all:', *(key for key in sorted(inserts)), sep='\n\t', end='\n\n') # copy newer modtimes of nested inserts to clients propagateModtimes(insmodtimes) # expand nested inserts replacements first expandInserts(inserts) # check run mode if len(sys.argv) == 1: filestoprocess = os.listdir(SOURCEDIR) # all files in source dir else: filestoprocess = sys.argv[1:] # or just filename(s) in args # expand and copy templates, copy others generateHtmls(filestoprocess, inserts, insmodtimes) # wrap up summary = '\nDone: %d generated, %d copied, %d skipped, %d failed.' print(summary % (numcnv, numcpy, numskip, numfail)) if sys.platform.startswith('win'): input('Press enter to close.') # retain shell if clicked