Source code for gavo.utils.texttricks

"""
Formatting, text manipulation, string constants, and such.
"""

#c Copyright 2008-2023, the GAVO project <gavo@ari.uni-heidelberg.de>
#c
#c This program is free software, covered by the GNU GPL.  See the
#c COPYING file in the source distribution.


import calendar
import datetime
import functools
import os
import quopri
import random
import re
import string
import time
from email import utils as emailutils

from gavo.utils.dachstypes import (Any, Filename, Generator,
	Optional, StrToStrMap, TextIO, Tuple, Union)

try:
	from astropy.coordinates import angle_formats
except ImportError:
	# old astropy
	from astropy.coordinates import angle_utilities as angle_formats # type: ignore

from gavo.utils import codetricks
from gavo.utils import misctricks
from gavo.utils.excs import SourceParseError, StructureError

floatRE = r"[+-]?(?:\d+\.?\d*|\.\d+)(?:[eE][+-]?\d+)?"
dateRE = re.compile("\d\d\d\d-\d\d-\d\d$")
datetimeRE = re.compile("\d\d\d\d-\d\d-\d\dT\d\d:\d\d:\d\dZ?$")
identifierPattern = re.compile("[A-Za-z_][A-Za-z0-9_]*$")
isoTimestampFmt = "%Y-%m-%dT%H:%M:%SZ"
isoTimestampFmtNoTZ = "%Y-%m-%dT%H:%M:%S"
entityrefPat = re.compile("&([^;])+;")
looksLikeURLPat = re.compile("[a-z]{2,5}://")


# file names that don't cause headaches in URLs and are otherwise reasonable
# benign (so, let's disallow shell metachars while we're at it).
_SAFE_FILENAME = re.compile("[,-:=@-Z_a-z{}~-]+$")

xmlEntities = {
		'lt': '<',
		'gt': '>',
		'amp': '&',
		'apos': "'",
		'quot': '"',
}


[docs]def formatSize(val: float, sf: int=1) -> str: """returns a human-friendly representation of a file size. """ if val<1e3: return "%d Bytes"%int(val) elif val<1e6: return "%.*fkiB"%(sf, val/1024.) elif val<1e9: return "%.*fMiB"%(sf, val/1024./1024.) else: return "%.*fGiB"%(sf, val/1024./1024./1024)
[docs]def makeEllipsis(aStr: str, maxLen: int=60, ellChars: str="...") -> str: """returns aStr cropped to maxLen if necessary. Cropped strings are returned with an ellipsis marker. """ aStr = re.sub("\s+", " ", aStr) if len(aStr)>maxLen: return aStr[:maxLen-len(ellChars)]+ellChars return aStr
[docs]def makeLeftEllipsis(aStr: str, maxLen: int=60): """returns aStr shortened to maxLen by dropping prefixes if necessary. Cropped strings are returned with an ellipsis marker. >>> makeLeftEllipsis("0123456789"*2, 11) '...23456789' """ aStr = re.sub("\s+", " ", aStr) if len(aStr)>maxLen: return "..."+aStr[-maxLen+3:] return aStr
[docs]def makeSourceEllipsis(sourceToken: Any) -> str: """returns a string hopefully representative for a source token. These are, in particular, passed around within rsc.makeData. Usually, these are (potentially long) strings, but now and then they can be other things with appallingly long reprs. When DaCHS messages need to refer to such sources, this function is used to come up with representative strings. """ if isinstance(sourceToken, str): return makeLeftEllipsis(sourceToken) else: return makeEllipsis(repr(sourceToken), maxLen=160)
[docs]@codetricks.document def getFileStem(fPath: str): """returns the file stem of a file path. The base name is what remains if you take the base name and split off extensions. The extension here starts with the last dot in the file name, except up to one of some common compression extensions (.gz, .xz, .bz2, .Z, .z) is stripped off the end if present before determining the extension. >>> getFileStem("/foo/bar/baz.x.y") 'baz.x' >>> getFileStem("/foo/bar/baz.x.gz") 'baz' >>> getFileStem("/foo/bar/baz") 'baz' """ for ext in [".gz", ".xz", ".bz2", ".Z", ".z"]: if fPath.endswith(ext): fPath = fPath[:-len(ext)] break return os.path.splitext(os.path.basename(fPath))[0]
[docs]def formatSimpleTable( data: list[list[Any]], stringify: bool=True, titles: Optional[list[str]]=None) -> str: """returns a string containing a text representation of tabular data. All columns of data are simply stringified, then the longest member determines the width of the text column. The behaviour if data does not contain rows of equal length is unspecified; data must contain at least one row. If you have serialised the values in data yourself, pass stringify=False. If you pass titles, it must be a sequence of strings; they are then used as table headers; the shorter of data[0] and titles will determine the number of columns displayed. """ if stringify: data = [[str(v) for v in row] for row in data] if not data: return "" colWidthes = [max(len(row[colInd]) for row in data) for colInd in range(len(data[0]))] if titles is not None: colWidthes = [max(len(t), l) for t, l in zip(titles, colWidthes)] fmtStr = " ".join("%%%ds"%w for w in colWidthes) table = "\n".join(fmtStr%tuple(row) for row in data) if titles is not None: table = fmtStr%tuple(titles)+"\n\n"+table return table
[docs]@codetricks.document def getRelativePath( fullPath: str, rootPath: str, liberalChars: bool=True) -> str: """returns rest if fullPath has the form rootPath/rest and raises an exception otherwise. Pass ``liberalChars=False`` to make this raise a ValueError when URL-dangerous characters (blanks, amperands, pluses, non-ASCII, and similar) are present in the result. This is mainly for products. """ if not fullPath.startswith(rootPath): raise ValueError( "Full path %s does not start with resource root %s"%(fullPath, rootPath)) res = fullPath[len(rootPath):].lstrip("/") if not liberalChars and not _SAFE_FILENAME.match(res): raise ValueError("File path '%s' contains characters known to" " the DaCHS authors to be hazardous in URLs. Please defuse the name" " before using it for published names (or see howDoI)."%res) return res
[docs]def resolvePath(rootPath: str, relPath: str) -> str: """joins relPath to rootPath and makes sure the result really is in rootPath. """ relPath = relPath.lstrip("/") fullPath = os.path.realpath(os.path.join(rootPath, relPath)) if not fullPath.startswith(rootPath): raise ValueError( "Full path %s does not start with resource root %s"%(fullPath, rootPath)) if not os.path.exists(fullPath): raise ValueError( "Invalid path %s. This should not happen."%(fullPath)) return fullPath
[docs]def fixIndentation(code: str, newIndent: str, governingLine: int=0) -> str: """returns code with all whitespace from governingLine removed from every line and newIndent prepended to every line. governingLine lets you select a line different from the first one for the determination of the leading white space. Lines before that line are left alone. >>> fixIndentation(" foo\\n bar", "") 'foo\\nbar' >>> fixIndentation(" foo\\n bar", " ") ' foo\\n bar' >>> fixIndentation(" foo\\n bar\\n baz", "", 1) 'foo\\nbar\\n baz' >>> fixIndentation(" foo\\nbar", "") Traceback (most recent call last): gavo.utils.excs.StructureError: Bad indent in line 'bar' """ codeLines = [line for line in code.split("\n")] reserved, codeLines = codeLines[:governingLine], codeLines[governingLine:] while codeLines: if codeLines[0].strip(): if mat := re.match("^\s*", codeLines[0]): firstIndent = mat.group() break else: reserved.append(codeLines.pop(0)) if codeLines: fixedLines = [] for line in codeLines: if not line.strip(): fixedLines.append(newIndent) else: if line[:len(firstIndent)]!=firstIndent: raise StructureError("Bad indent in line %s"%repr(line)) fixedLines.append(newIndent+line[len(firstIndent):]) else: fixedLines = codeLines reserved = [newIndent+l.lstrip() for l in reserved] return "\n".join(reserved+fixedLines)
@functools.lru_cache() def _getREForPercentExpression(format: str) -> re.Pattern: """helps parsePercentExpression. """ parts = re.split(r"(%\w)", format) newReParts = [] for ind, p in enumerate(parts): if p.startswith("%"): # the time-parsing hack explained in the docstring: if ind+2<len(parts) and parts[ind+1]=="": if p[1] in "HMS": newReParts.append("(?P<%s>..)"%p[1]) else: raise ValueError( "At %s: conversions with no intervening literal not supported."% p) else: newReParts.append("(?P<%s>.*?)"%p[1]) else: newReParts.append(re.escape(p)) return re.compile("".join(newReParts)+"$")
[docs]def parsePercentExpression(literal: str, format: str) -> dict: """returns a dictionary of parts in the %-template format. format is a template with %<conv> conversions, no modifiers are allowed. Each conversion is allowed to contain zero or more characters matched stingily. Successive conversions without intervening literals aren't really supported. There's a hack for strptime-type times, though: H, M, and S just eat two characters each if there's no separator. This is really only meant as a quick hack to support times like 25:33. >>> r=parsePercentExpression("12,xy:33,","%a:%b,%c"); r["a"], r["b"], r["c"] ('12,xy', '33', '') >>> sorted(parsePercentExpression("2357-x", "%H%M-%u").items()) [('H', '23'), ('M', '57'), ('u', 'x')] >>> r = parsePercentExpression("12,13,14", "%a:%b,%c") Traceback (most recent call last): ValueError: '12,13,14' cannot be parsed using format '%a:%b,%c' """ mat = _getREForPercentExpression(format).match(literal) if not mat: raise ValueError("'%s' cannot be parsed using format '%s'"%( literal, format)) return mat.groupdict()
[docs]def parseAssignments(assignments: str) -> StrToStrMap: """returns a name mapping dictionary from a list of assignments. This is the preferred form of communicating a mapping from external names to field names in records to macros -- in a string that contains ":"-seprated pairs separated by whitespace, like "a:b b:c", where the incoming names are leading, the desired names are trailing. If you need defaults to kick in when the incoming data is None, try _parseDestWithDefault in the client function. This function parses a dictionary mapping original names to desired names. >>> parseAssignments("a:b b:c") {'a': 'b', 'b': 'c'} """ return dict([(lead, trail) for lead, trail in [litPair.split(":") for litPair in assignments.split()]])
[docs]@codetricks.document def hmsToDeg(hms: str, sepChar: Optional[str]=None) -> float: """returns the time angle (h m s.decimals) as a float in degrees. >>> "%3.8f"%hmsToDeg("22 23 23.3") '335.84708333' >>> "%3.8f"%hmsToDeg("22:23:23.3", ":") '335.84708333' >>> "%3.8f"%hmsToDeg("222323.3", "") '335.84708333' >>> hmsToDeg("junk") Traceback (most recent call last): ValueError: Invalid time with sepChar None: 'junk' """ hms = hms.strip() try: if sepChar=="": parts = [hms[:2], hms[2:4], hms[4:]] else: parts = hms.split(sepChar) if len(parts)==3: hours, minutes, seconds = parts elif len(parts)==2: hours, minutes = parts seconds = "0" else: raise ValueError("Too many parts") timeSeconds = int(hours)*3600+float(minutes)*60+float(seconds or "0") except ValueError: raise ValueError("Invalid time with sepChar %s: %s"%( repr(sepChar), repr(hms))) return timeSeconds/3600/24*360
[docs]@codetricks.document def dmsToDeg(dmsAngle: str, sepChar: Optional[str]=None) -> float: """returns the degree minutes seconds-specified dmsAngle as a float in degrees. >>> "%3.8f"%dmsToDeg("45 30.6") '45.51000000' >>> "%3.8f"%dmsToDeg("45:30.6", ":") '45.51000000' >>> "%3.8f"%dmsToDeg("-45 30 7.6") '-45.50211111' >>> dmsToDeg("junk") Traceback (most recent call last): ValueError: Invalid dms value with sepChar None: 'junk' """ dmsAngle = dmsAngle.strip() sign = 1 if dmsAngle.startswith("+"): dmsAngle = dmsAngle[1:].strip() elif dmsAngle.startswith("-"): sign, dmsAngle = -1, dmsAngle[1:].strip() try: if sepChar=="": parts = [dmsAngle[:2], dmsAngle[2:4], dmsAngle[4:]] else: parts = dmsAngle.split(sepChar) if len(parts)==3: deg, min, sec = parts elif len(parts)==2: deg, min = parts sec = "0" else: raise ValueError("Invalid # of parts") arcSecs = sign*(int(deg)*3600+float(min)*60+float(sec or "0")) except ValueError: raise misctricks.logOldExc( ValueError("Invalid dms value with sepChar %s: %s"%( repr(sepChar), repr(dmsAngle)))) return arcSecs/3600
[docs]def fracHoursToDeg(fracHours: float) -> float: """returns the time angle fracHours given in decimal hours in degrees. """ return float(fracHours)*360./24.
[docs]@codetricks.document def degToHms(deg: float, sepChar: str=" ", secondFracs: int=3, truncate: bool=False) -> str: """converts a float angle in degrees to an time angle (hh:mm:ss.mmm). This takes a lot of optional arguments: * sepChar is the char separating the components * secondFracs is the number for fractional seconds to generate * truncate can be set to True if fractional seconds should be truncated rather then rounded (as necessary for building IAU identifiers) >>> degToHms(0, sepChar=":") '00:00:00.000' >>> degToHms(122.057, secondFracs=1) '08 08 13.7' >>> degToHms(122.057, secondFracs=1, truncate=True) '08 08 13.6' >>> degToHms(-0.055, secondFracs=0) '-00 00 13' >>> degToHms(-0.055, secondFracs=0, truncate=True) '-00 00 13' >>> degToHms(-1.056, secondFracs=0) '-00 04 13' >>> degToHms(-1.056, secondFracs=0) '-00 04 13' >>> degToHms(359.9999999) '24 00 00.000' >>> degToHms(359.2222, secondFracs=4, sepChar=":") '23:56:53.3280' >>> "%.4f"%hmsToDeg(degToHms(256.25, secondFracs=9)) '256.2500' """ if truncate: return angle_formats.hours_to_string( deg/360*24, sep=(sepChar, sepChar), precision=secondFracs+8, pad=True)[:-8].rstrip(".") return angle_formats.hours_to_string( deg/360*24, sep=(sepChar, sepChar), precision=secondFracs, pad=True)
[docs]@codetricks.document def hoursToHms(decimal_hours: float, sepChar: str=":", secondFracs:int =0) -> str: """returns a time span in hours in sexagesmal time (h:m:s). The optional arguments are as for degToHms. >>> hoursToHms(0) '00:00:00' >>> hoursToHms(23.5) '23:30:00' >>> hoursToHms(23.55) '23:33:00' >>> hoursToHms(23.525) '23:31:30' >>> hoursToHms(23.553, secondFracs=2) '23:33:10.80' >>> hoursToHms(123.553, secondFracs=2) '123:33:10.80' """ return angle_formats.hours_to_string( decimal_hours, sep=(sepChar, sepChar), precision=secondFracs, pad=True)
[docs]@codetricks.document def degToDms(deg: float, sepChar: str=" ", secondFracs: int=2, preserveLeading: bool=False, truncate: bool=False, addSign: bool=True) -> str: """converts a float angle in degrees to a sexagesimal string. This takes a lot of optional arguments: * sepChar is the char separating the components * secondFracs is the number for fractional seconds to generate * preserveLeading can be set to True if leading zeroes should be preserved * truncate can be set to True if fractional seconds should be truncated rather then rounded (as necessary for building IAU identifiers) * addSign, if true, makes the function return a + in front of positive values (the default) >>> degToDms(-3.24722, "", 0, True, True) '-031449' >>> degToDms(0) '+0 00 00.00' >>> degToDms(0, addSign=False) '0 00 00.00' >>> degToDms(-0.25, sepChar=":") '-0:15:00.00' >>> degToDms(-23.50, secondFracs=4) '-23 30 00.0000' >>> "%.4f"%dmsToDeg(degToDms(-25.6835, sepChar=":"), sepChar=":") '-25.6835' """ if truncate: secondFracs += 8 fmted = angle_formats.degrees_to_string( deg, sep=(sepChar, sepChar), precision=secondFracs, pad=preserveLeading) if truncate: fmted = fmted[:-8].rstrip(".") if addSign and deg>=0: fmted = "+"+fmted return fmted
[docs]@codetricks.document def makeIAUId(prefix: str, long: float, lat: float, longSec: int=0, latSec: int=0) -> str: """returns an (equatorial) IAU identifier for an object at long and lat. The rules are given on https://cds.unistra.fr/Dic/iau-spec.html The prefix, including the system identifier, you have to pass in. You cannot build identifiers using only minutes precision. If you want to include sub-arcsec precision, pass in longSec and/or latSec (the number of factional seconds to preserve). """ longPart = degToHms(long, "", longSec, True) latPart = degToDms(lat, "", latSec, True, True) return f"{prefix}{longPart}{latPart}"
[docs]def datetimeToRFC2616(dt: datetime.datetime) -> str: """returns a UTC datetime object in the format required by http. This may crap when you fuzz with the locale. In general, when handling "real" times within the DC, prefer unix timestamps over datetimes and use the other ``*RFC2616`` functions. """ return dt.strftime('%a, %d %b %Y %H:%M:%S GMT')
[docs]def parseRFC2616Date(s: str) -> float: """returns seconds since unix epoch representing UTC from the HTTP-compatible time specification s. """ parts = emailutils.parsedate_tz(s) assert parts is not None return emailutils.mktime_tz(parts)
# The following timegm implementation is due to Frederik Lundh def _d(y: int, m: int, d: int, days: Tuple[int, ...]=(0,31,59,90,120,151,181,212,243,273,304,334,365) ) -> int: return (((y - 1901)*1461)//4 + days[m-1] + d + ( (m > 2 and not y % 4 and (y % 100 or not y % 400)) and 1))
[docs]def timegm(tm: time.struct_time, epoch: float=_d(1970,1,1)): year, month, day, h, m, s = tm[:6] return (_d(year, month, day) - epoch)*86400 + h*3600 + m*60 + s
[docs]def formatRFC2616Date(secs: Optional[float]=None) -> str: """returns an RFC2616 date string for UTC seconds since unix epoch. """ if secs is None: secs = time.time() return emailutils.formatdate(secs, localtime=False, usegmt=True)
_isoDTRE = re.compile(r"(?P<year>\d\d\d\d)-?(?P<month>\d\d)-?(?P<day>\d\d)" r"(?:[T ](?P<hour>\d\d):?(?P<minute>\d\d):?" r"(?P<seconds>\d\d)(?P<secFracs>\.\d*)?Z?(\+00:00)?)?$")
[docs]@codetricks.document def parseISODT(literal: str, useTime: bool=False) -> datetime.datetime: """returns a datetime object for a ISO time literal. There's no real timezone support yet, but we accept and ignore various ways of specifying UTC. By default, this uses plain python datetime because it usually covers a large date range than the time module. The downside is that it does not know about leap seconds. Pass useTime=True to go through time tuples, which know how to deal with them (but may not deal with dates far in the past or future). >>> parseISODT("1998-12-14") datetime.datetime(1998, 12, 14, 0, 0) >>> parseISODT("1998-12-14T13:30:12") datetime.datetime(1998, 12, 14, 13, 30, 12) >>> parseISODT("1998-12-14T13:30:12Z") datetime.datetime(1998, 12, 14, 13, 30, 12) >>> parseISODT("1998-12-14T13:30:12.224Z") datetime.datetime(1998, 12, 14, 13, 30, 12, 224000) >>> parseISODT("19981214T133012Z") datetime.datetime(1998, 12, 14, 13, 30, 12) >>> parseISODT("19981214T133012+00:00") datetime.datetime(1998, 12, 14, 13, 30, 12) >>> parseISODT("2016-12-31T23:59:60") Traceback (most recent call last): ValueError: second must be in 0..59 >>> parseISODT("2016-12-31T23:59:60", useTime=True) datetime.datetime(2017, 1, 1, 1, 0) >>> parseISODT("junk") Traceback (most recent call last): ValueError: Bad ISO datetime literal: junk (required format: yyyy-mm-ddThh:mm:ssZ) """ if isinstance(literal, datetime.datetime): return literal literal = literal.rstrip("Z") mat = _isoDTRE.match(literal.strip()) if not mat: raise ValueError("Bad ISO datetime literal: %s" " (required format: yyyy-mm-ddThh:mm:ssZ)"%literal) parts = mat.groupdict() if parts["hour"] is None: parts["hour"] = parts["minute"] = parts["seconds"] = 0 if parts["secFracs"] is None: parts["secFracs"] = 0 else: parts["secFracs"] = "0"+parts["secFracs"] if useTime: return datetime.datetime.fromtimestamp( calendar.timegm(( # type calculus broken for RE results in mypy 1.0.1 int(parts["year"]), int(parts["month"]), int(parts["day"]), # type: ignore int(parts["hour"]), int(parts["minute"]), int(parts["seconds"]), # type: ignore -1, -1, -1))) else: return datetime.datetime(int(parts["year"]), int(parts["month"]), # type: ignore int(parts["day"]), int(parts["hour"]), int(parts["minute"]), # type: ignore int(parts["seconds"]), int(float(parts["secFracs"])*1000000)) # type: ignore
_SUPPORTED_DT_FORMATS =[ '%Y-%m-%dT%H:%M:%S', '%Y-%m-%d %H:%M:%S', '%Y-%m-%d',]
[docs]def parseDefaultDatetime(literal: Optional[Union[str, datetime.datetime]] ) -> Optional[datetime.datetime]: """returns a datetime from string or passes through datetimes and Nones. The function will try to parse a string in various ways; we will try not to drop formats from one minor version to the next. """ if literal is None or isinstance(literal, datetime.datetime): return literal if literal.endswith("Z"): literal = literal[:-1] for format in _SUPPORTED_DT_FORMATS: try: return datetime.datetime( *time.strptime(literal, format)[:6]) except ValueError: pass return parseISODT(literal)
[docs]def parseDefaultDate(literal: Optional[Union[str, datetime.date]] ) -> Optional[datetime.date]: """parseDefaultDatetime's little sister. """ if literal is None or isinstance(literal, datetime.date): return literal return datetime.date(*time.strptime(literal, '%Y-%m-%d')[:3])
[docs]def parseDefaultTime(literal: Optional[Union[str, datetime.time]] ) -> Optional[datetime.time]: """parseDefaultDatetime's other little sister. """ if literal is None or isinstance(literal, datetime.time): return literal # as long as we're builing on top of time, we can't do fractional seconds return datetime.time(*time.strptime(literal, '%H:%M:%S')[3:6])
[docs]def roundToSeconds(dt: datetime.datetime) -> datetime.datetime: """returns a datetime instance rounded to whole seconds. This also recklessly clears any time zone marker. So, don't pass in anything with a meaningful time zone. """ if dt.microsecond>500000: return dt.replace(microsecond=0, tzinfo=None )+datetime.timedelta(seconds=1) else: return dt.replace(microsecond=0, tzinfo=None)
[docs]def formatISODT(dt: datetime.datetime) -> str: """returns some ISO8601 representation of a datetime instance. The reason for preferring this function over a simple str is that datetime's default representation is too difficult for some other code (e.g., itself); hence, this code suppresses any microsecond part and always adds a Z (where strftime works, utils.isoTimestampFmt produces an identical string). The behaviour of this function for timezone-aware datetimes is undefined. For convenience, None is returned as None. Also for convenience, you can pass in a string; this will then be parsed first, which provides both some basic format validation and guaranteed DALI-compliant serialisation. >>> formatISODT(datetime.datetime(2015, 10, 20, 12, 34, 22, 250)) '2015-10-20T12:34:22Z' >>> formatISODT(datetime.datetime(1815, 10, 20, 12, 34, 22, 250)) '1815-10-20T12:34:22Z' >>> formatISODT(datetime.datetime(2018, 9, 21, 23, 59, 59, 640000)) '2018-09-22T00:00:00Z' """ if dt is None: return None if isinstance(dt, str): dt = parseDefaultDatetime(dt) return roundToSeconds(dt).isoformat()+"Z"
[docs]@codetricks.document def formatFloat(f: float) -> str: """returns floating-point numbers somewhat suitable for human consumption. The idea of this function is to slowly migrate ad-hoc formatting of this kind that's in all kind of different places in DaCHS here and thus have a central knob we can eventually use to adapt to tastes. >>> formatFloat(1) '1' >>> formatFloat(1/3) '0.333333' >>> formatFloat(-1/3e20) '-3.33333e-21' >>> import math;formatFloat(math.pi) '3.14159' >>> formatFloat(20000000.23) '2e+07' """ return "{:.6g}".format(f)
[docs]class NameMap(object): """is a name mapper fed from a simple text file. The text file format simply is: <target-id> "TAB" <src-id>{whitespace <src-id>} src-ids have to be encoded quoted-printable when they contain whitespace or other "bad" characters ("="!). You can have #-comments and empty lines. The file is supposed to be ASCII, with non-ASCII encoded quoted-printable. The qp-decoded strings are assumed to be utf-8 encoded, but there's a constructor argument to change that. """ def __init__(self, src: Filename, missingOk: bool=False, enc: str="utf-8"): self._parseSrc(src, missingOk, enc) def __contains__(self, name: str) -> bool: return name in self.namesDict def _parseSrc(self, src: Filename, missingOk: bool, enc: str) -> None: self.namesDict: StrToStrMap = {} try: f = open(src, "rb") except IOError: if not missingOk: raise else: return try: for ln in f: if ln.startswith(b"#") or not ln.strip(): continue ob, names = re.split(b"\t+", ln) for name in names.lower().split(): self.namesDict[quopri.decodestring(name).decode(enc) ] = ob.decode(enc) except ValueError: raise misctricks.logOldExc(ValueError( "Syntax error in %s: Line %s not understood."%(src, repr(ln)))) f.close()
[docs] def resolve(self, name: str) -> str: return self.namesDict[name.lower()]
_STANDARD_ENTITIES = { 'lt': '<', 'gt': '>', 'amp': '&', 'apos': "'", 'quot': '"', } def _decodeEntityref(matob: re.Match) -> str: entRef = matob.group(1) if entRef in _STANDARD_ENTITIES: return _STANDARD_ENTITIES[entRef] elif entRef.startswith("#x"): return chr(int(entRef[2:], 16)) elif entRef.startswith("#"): return chr(int(entRef[1:])) else: raise ValueError("Unknown entity reference: &%s;"%entRef)
[docs]def replaceXMLEntityRefs(unicodeString: str) -> str: """retplaces all known HTML entities in unicodeString with actual unicode chars. (and dies on unknown entities). TODO: this is unused and probably not very useful to clients. Discard? """ return entityrefPat.sub(_decodeEntityref, unicodeString)
[docs]def ensureOneSlash(s: str) -> str: """returns s with exactly one trailing slash. """ return s.rstrip("/")+"/"
_SimpleTextGenerator = Generator[Tuple[int, str], None, None] def _iterSimpleTextNoContinuation(f: TextIO ) -> _SimpleTextGenerator: """helps iterSimpleText. """ for (lineNumber, curLine) in enumerate(f): curLine = curLine.strip() if curLine and not curLine.startswith("#"): yield (lineNumber+1), curLine
[docs]@codetricks.document def iterSimpleText(f: TextIO) -> _SimpleTextGenerator: """iterates over ``(physLineNumber, line)`` in f with some usual conventions for simple data files. You should use this function to read from simple configuration and/or table files that don't warrant a full-blown grammar/rowmaker combo. The intended use is somewhat like this:: with open(rd.getAbsPath("res/mymeta")) as f: for lineNumber, content in iterSimpleText(f): try: ... except Exception, exc: sys.stderr.write("Bad input line %s: %s"%(lineNumber, exc)) The grammar rules are, specifically: * leading and trailing whitespace is stripped * empty lines are ignored * lines beginning with a hash are ignored * lines ending with a backslash are joined with the following line; to have intervening whitespace, have a blank in front of the backslash. """ iter = _iterSimpleTextNoContinuation(f) try: while True: lineNumber, curLine = next(iter) while curLine.endswith("\\"): try: lineNumber, newStuff = next(iter) except StopIteration: raise SourceParseError("File ends with a backslash", location="line %d"%lineNumber) curLine = curLine[:-1]+newStuff yield lineNumber, curLine except StopIteration: # all done, leave loop pass
_RANDOM_STRING_OK_CHARS = string.ascii_letters+string.digits+"_.,"
[docs]def getRandomString(length: int) -> str: """returns a random string of harmless printable characters. """ return "".join( random.choice(_RANDOM_STRING_OK_CHARS) for c in range(length))
[docs]def safe_str(val: Any) -> str: """returns a reasonable string from pretty much anything. """ if isinstance(val, str): return val elif isinstance(val, bytes): return val.decode("ascii", "ignore") else: return str(val)
[docs]def bytify(s: Union[str, bytes]) -> bytes: """returns s utf-8 encoded if it is a string, unmodified otherwise. """ if isinstance(s, str): return s.encode("utf-8") return s
[docs]def debytify(b: Union[str, bytes], enc: str="ascii"): """returns a bytestring b as a normal string. This will return b unless it's bytes. If it is bytes, it will be decoded as enc (and the thing will fail when the encoding is wrong). """ if isinstance(b, bytes): return b.decode(enc, "ignore") return b
DEFUSE_NONPRINTABLE = bytes.maketrans(bytes(range(32)), b" "*32)
[docs]def defuseFileName(fName: Union[Filename, bytes], replaceSlash: bool=True): """returns fName without any non-ASCII or slashes but in a way that people can still work out what the file name has been. This is basically a quoted-printable encoding. What's returned is a string that's guaranteed to be ASCII only. With replaceSlash=False, it can also double as a reasonable asciification. """ nameBytes = fName.encode("utf-8") if isinstance(fName, str) else fName res = quopri.encodestring(nameBytes ).translate(DEFUSE_NONPRINTABLE ).decode("ascii") if replaceSlash: res = res.replace('/', "=2F") return res
[docs]def parseAccept(aString: str) -> StrToStrMap: """parses an RFC 2616 accept header and returns a dict mapping media type patterns to their (unparsed) parameters. If aString is None, an empty dict is returned If we ever want to do fancy things with http content negotiation, this will be further wrapped to provide something implementing the complex RFC 2616 rules; this primitive interface really is intended for telling apart browsers (which accept text/html) from other clients (which hopefully do not) at this point. >>> sorted(parseAccept("text/html, text/*; q=0.2; level=3").items()) [('text/*', 'q=0.2; level=3'), ('text/html', '')] >>> parseAccept(None) {} """ res = {} if aString is not None: for item in aString.split(","): if ";" in item: key, params = item.split(";", 1) else: key, params = item, "" res[key.strip()] = params.strip() return res
if __name__=="__main__": # pragma: no cover import doctest doctest.testmod()