#!/usr/bin/python3
"""sscDD.py -- Python port of sscDD.php v1.1.2 (CFBaum).

Reads the BOC bocode RDF files (x and y) from RePEc and prints one line per
Stata package:  " <name> <date>", where <date> is the Revision-Date if it is
later than the Creation-Date, otherwise the Creation-Date.

Writes the listing to standard output; run from cron and redirect it to the
static file that Stata reads. Exits with status 1 (and writes nothing to
standard output) if either source file cannot be fetched.

Notes on fidelity to the PHP original:
  * Lines are handled as raw bytes and split on "\n" only, as PHP file() does,
    so non-UTF-8 characters and stray "\r" pass through unchanged.
  * The regexes have the same semantics as the PHP preg_match patterns
    ("$" also matches before a final newline; "." does not match "\n").
  * The PHP fmt() function was dead code (its results were never used),
    so it is not ported.
  * The date comparison mirrors PHP 8's "$revd > $cred": numeric when both
    are numeric strings, otherwise a plain string comparison (so an empty
    Revision-Date never wins).

cfb 20261025: removed bocodey
"""
import io
import sys
import re
import urllib.request

SOURCES = (
    "http://RePEc.org/boc/bocode/bocodex.rdf",
)
TIMEOUT = 30  # seconds per fetch

TITLE_RE = re.compile(rb"^Title: (.+): Stata (.+)$")
CREATED_RE = re.compile(rb"^Creation-Date: (.+)$")
REVISED_RE = re.compile(rb"^Revision-Date: (.+)$")
HANDLE_RE = re.compile(rb"^Handle: (.+)bocode:S(.+)\n$")

CHOP = b" \t\n\r\0\x0b"  # characters stripped by PHP chop()/rtrim()


def is_later(a, b):
    """Return PHP 8's result for `a > b` on two date strings."""
    if a.isdigit() and b.isdigit():
        return int(a) > int(b)
    return a > b


def fetch_lines():
    """Fetch every source file and return their lines, in order."""
    lines = []
    for url in SOURCES:
        with urllib.request.urlopen(url, timeout=TIMEOUT) as resp:
            lines.extend(io.BytesIO(resp.read()).readlines())
    return lines


def build_listing(lines):
    """Return the listing (bytes) for the given RDF lines."""
    out = []
    name = b""
    cred = b""
    revd = b""
    for line in lines:
        m = TITLE_RE.match(line)
        if m:
            name = m.group(1)
        m = CREATED_RE.match(line)
        if m:
            cred = m.group(1).rstrip(CHOP)
        m = REVISED_RE.match(line)
        if m:
            revd = m.group(1).rstrip(CHOP)
        if HANDLE_RE.match(line):
            date = revd if is_later(revd, cred) else cred
            out.append(b" " + name + b" " + date + b"\n")
            cred = b""
            revd = b""
    return b"".join(out)


def main():
    try:
        body = build_listing(fetch_lines())
    except Exception as exc:  # network or HTTP error reading RePEc
        sys.stderr.write("sscDD.py: could not read source data: %s\n" % exc)
        sys.exit(1)  # nonzero, so "&& mv" in cron keeps the old file
    out = sys.stdout.buffer
    out.write(body)
    out.flush()


if __name__ == "__main__":
    main()