#! /usr/bin/python3
# Last edited on 2026-01-22 11:42:30 by stolfi

# Library function that reads the parags property table of the SPS.
# To be imported by python programs.

from process_funcs import basic_line_loop
import re, sys
from sys import stderr as err

def read_table(fname):
  # Reads the contents of the parags property from file {fname}. 
  # 
  # Blank lines and #-comments in file {fname} are ignored.
  #
  # For each page there must be one line starting with "PLOC" that has the header for the 
  # parag table of that page summaries.  
  #
  # Then for each parag in that page there must be a line
  # describing that parag, which must have the format
  # 
  #   "{hloc} | P{nn} | {props}"
  # 
  # where 
  # 
  #   {hloc} is "{fnum}.{hlseq}".
  #     {fnum} is the page's f-number.
  #     {hlseq} is the line seq number {lseq} of the parag's head line.
  #   {nn} is the seq number {pseq} of the parag in the page as 2 digits.
  #   {props} are the parag propertis.
  #
  # There must be one line starting with "PAGE" that has the header for the 
  # page summary.  
  #
  # Then there must be a page summary line with various counts
  # for that page, such as number of lines, stars, parags, perfect parags, etc.
  # This line must have the format 
  # 
  #   "{fnum} | TOT | {stats}" 
  # 
  # where {fnum} is a page's f-number and {stats} are the page statistics, in any format
  #
  # At the end of the file there should be a file summary line that is
  # like the page summary line but has "TOTAL" instead of the {fnum},
  # and whose statistics are the totals for all pages.
  # 
  # Returns
  # 
  #   {hdr_page} header for the page summaries
  #   {tbl_page} table of page sumaries {stats}, indexed by {[fnum]}.
  #   {hdr_parag} header for the per-page parags table.
  #   {tbl_parag} table of parag properties {props}, indexed by {[fnum][pseq-1]}
  #
  # For now, the {stats} and {props} are unparsed strings.
  
  page_tbl = dict()
  page_hdr = None
  fnum_cur = None
  
  parag_hdr = None
  parag_tbl = dict()
  last_pseq = None
  
  def process_line(nread, line):
    nonlocal page_tbl, page_hdr, fnum_cur, parag_hdr, parag_tbl, last_pseq
    line = line.strip()
    if re.match(r"^[ \011]*([#]|$)", line): return
    # Parse parag properties header:
    if re.match(r"PLOC ", line):
      parag_hdr = line
      return
    # Parse parag properties line:
    m = re.fullmatch(r"(f[0-9a-z]+[.][0-9]+)[ ]*[|][ ]*(P[0-9][0-9])[ ]*[|][ ](.*)", line)
    if m != None:
      locid, Pnn, props = m.group(1,2,3)
      fnum= re.sub(r"[.][0-9]+$", "", locid)
      if fnum_cur == "TOTAL":
        err.write(f"{fname}:{nread}: ** parags after TOTAL\n  [[{line}]]\n")
        sys.exit(1)
      if fnum != fnum_cur:
        err.write(f"!! start of page {fnum}\n")
        if fnum in parag_tbl:
          err.write(f"{fname}:{nread}: ** repeated page = {fnum}\n  [[{line}]]\n")
          sys.exit(1)
        parag_tbl[fnum] = []
        fnum_cur = fnum
        last_pseq = 0
      pseq = int(Pnn[1:])
      if pseq != last_pseq + 1:
        err.write(f"{fname}:{nread}: ** unexpected parag number = {Pnn} {last_pseq = }\n  [[{line}]]\n")
        sys.exit(1)
      parag_tbl[fnum].append(line)
      last_pseq = pseq
      return
    # Parse page summary header:
    if re.match(r"PAGE ", line):
      page_hdr = line
      return
    # Parse page summary line:
    m = re.match(r"(TOTAL|f[0-9a-z]+)[ ]*[|][ ](.*) ", line)
    if m != None:
      fnum, props = m.group(1,2)
      if fnum_cur == "TOTAL":
        err.write(f"{fname}:{nread}: ** page summary after TOTAL\n  [[{line}]]\n")
        sys.exit(1)
      if fnum != "TOTAL" and fnum != fnum_cur:
        err.write(f"{fname}:{nread}: ** summary for wrong page {fnum} {fnum_cur = }\n  [[{line}]]\n")
        sys.exit(1)
      if fnum in page_tbl:
        err.write(f"{fname}:{nread}: ** repeated page summary {fnum}\n  [[{line}]]\n")
        sys.exit(1)
      page_tbl[fnum] = line
      return
    err.write(f"{fname}:{nread}: ** invalid line format\n  [[{line}]]\n")
    sys.exit(1)
    # return
    # ....................................................................
    
  rd = open(fname, "r")
  nread = basic_line_loop(rd, process_line)
  if len(page_tbl) == 0 or page_hdr == None:
    err.write(f"{fname}:{nread}: ** no page summaries in table file {fname}\n")
    sys.exit(1)
  if "TOTAL" not in page_tbl:
    err.write(f"{fname}:{nread}: ** no TOTAL summary in table file {fname}\n")
    sys.exit(1)
  if len(parag_tbl) == 0 or parag_hdr == None:
    err.write(f"{fname}:{nread}: ** no parag entries in table file {fname}\n")
    sys.exit(1)
  rd.close()
  return page_hdr, page_tbl, parag_hdr, parag_tbl
  # ----------------------------------------------------------------------