# Puts the no-break spaces French typography wants into the French books:  before ; : ! ? and
# inside guillemets.  A translator types ordinary spaces;  this makes them U+00A0 in prose only,
# leaving listings, inline code, attribute and block attribute lines, comments, and AsciiDoc's own
# colons (term::, NOTE:, macros) alone.  Run it after editing docs/fr, then util/checkTranslations.rb.
#
# Usage:  python3 util/frenchSpaces.py [file...]       every .adoc under docs/fr, or the files named
import glob
import os
import re
import sys


REPO   = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
NBSP   = "\u00a0"
FENCE  = re.compile(r"^(-{4,}|\.{4,}|\+{4,}|/{4,})$")
BEFORE = re.compile("(?<=\\S) ([;:!?\u00bb])(?=\\s|$|[\u00bb)\\]*_,.])")
AFTER  = re.compile("\u00ab ")
CODE   = re.compile(r"`[^`\n]*`|\b[a-z]+:\S*\[")
LINK   = re.compile(r"https?://[^\s\[\]<>]+")
HOLD   = "\x00%d\x00"
LINKED = "\x01%d\x01"


def fixProse(line):
	# Inline code, web addresses, and macros are set aside behind placeholders while the spaces are
	# fixed, so a space after one (`DEBUG_SHAPES` :) is still seen as following a word.
	kept  = []

	def keep(match):
		kept.append(match.group(0))
		return HOLD % (len(kept) - 1)

	def keepLink(match):
		kept.append(match.group(0))
		return LINKED % (len(kept) - 1)

	# A web address ends at the first space, and Asciidoctor takes a no-break space as part of it,
	# so the space after one stays ordinary.
	fixed = fixStretch(CODE.sub(keep, LINK.sub(keepLink, line)))
	fixed = re.sub(r"(\x01\d+\x01)" + NBSP, r"\1 ", fixed)
	return re.sub(r"[\x00\x01](\d+)[\x00\x01]", lambda m: kept[int(m.group(1))], fixed)


def fixStretch(text):
	# A colon followed by another is AsciiDoc's (term::), never French punctuation.
	text = BEFORE.sub(lambda m: m.group(0) if (m.group(1) == ":" and text[m.end():m.end() + 1] == ":") else NBSP + m.group(1), text)
	return AFTER.sub("\u00ab" + NBSP, text)


def fixFile(path):
	lines   = open(path, encoding="utf-8").read().split("\n")
	fence   = None
	changed = 0
	for index, line in enumerate(lines):
		if FENCE.match(line):
			fence = None if fence == line else (fence or line)
			continue
		if fence or line.startswith((":", "[", "//", "include::", "image::")) or line.startswith("|==="):
			continue
		fixed = fixProse(line)
		if fixed != line:
			lines[index] = fixed
			changed += 1
	if changed:
		open(path, "w", encoding="utf-8").write("\n".join(lines))
	return changed


def main():
	paths = sys.argv[1:] or sorted(glob.glob(os.path.join(REPO, "docs", "fr", "**", "*.adoc"), recursive=True))
	for path in paths:
		changed = fixFile(path)
		if changed:
			print("%s:  %d lines" % (os.path.relpath(path, REPO), changed))


main()