#! /usr/bin/python3
# Last edited on 2026-09-13 16:53:08 by stolfi

import sys, os, re;
from sys import stderr as err
from error_funcs import arg_error, prog_error
from process_funcs import bash
import bencao_starps_make_funcs as makef
import bencao_starps_size_position_funcs as spf
 
def add_all_071_rules(pre, mak, tit):
  # Adds rules to build all derived files that
  # depend only on the SPS sources.
  #
  # Returns the list of ultimate targets.
  
  targets = \
    add_all_071_bencao_par_ivt_rules(pre, mak, tit, "wh") + \
    add_all_071_bencao_par_ivt_rules(pre, mak, tit, "ch") + \
    add_all_071_upp_rules(pre, mak, tit) + \
    add_all_071_word_tuple_rules(pre, mak, tit) + \
    add_all_071_lines_with_pattern_rules(pre, mak, tit) + \
    add_all_071_single_hist_rules(pre, mak, tit) + \
    add_all_071_double_hist_rules(pre, mak, tit) + \
    add_all_071_bencao_word_pos_plot_rules(pre, mak, tit)

  return targets
  # ----------------------------------------------------------------------
  
def add_all_071_bencao_par_ivt_rules(pre, mak, tit, utype):
  # Rules to create the "bencao-fu-{utype}-par.ivt" file,
  # where {utype} is "ch" and "wh", 
  # from the raw transcription files in the "clean/ch" folder,
  # falling back to "ctp/bencao.uts" for entries not found there.

  targets = []

  source_master = "ctp/bencao.utf"
  clean_dir = "clean/ch"

  ivt_par_name = f"bencao-fu-{utype}-par"

  target = f"{ivt_par_name}.ivt"
  script = "create_bencao_fu_par_ivt.py"

  tit[target] = f"making res/{target} from {source_master}, {clean_dir}/*.utf"
  pre[target] = [ source_master, clean_dir, ]
  mak[target] = (
    f"./{script} {utype} {clean_dir} {source_master} res/{target}",
  )
  targets.append(target)

  return targets
  # ----------------------------------------------------------------------
  
def add_all_071_bencao_ch_par_ivt_rules(pre, mak, tit):
  # Rules to create "bencao-fu-ch-par.ivt"
  # from "bencao-fu-wh-par.ivt".

  targets = []

  wh_ivt_par_name = f"bencao-fu-wh-par"
  ch_ivt_par_name = f"bencao-fu-ch-par"

  source_wh = f"res/{wh_ivt_par_name}.ivt"

  target = f"{ch_ivt_par_name}.ivt"
  script = "remove_chinese_punctuation.py"

  tit[target] = f"making res/{target} from {source_wh}"
  pre[target] = [ source_wh, ]
  mak[target] = (
    f"./{script} < {source_wh} > res/{target}",
  )
  targets.append(target)

  return targets
  # ----------------------------------------------------------------------

def add_all_071_upp_rules(pre, mak, tit):
  # Rules to create the "{book}-{bsub}-{utype}.upp" files, with the size of
  # each recipe. The {utype} specifies the unit of size, namely "ch" for hanzi
  # chars.

  targets = []

  for book, bsub, utype, ltype in ( \
      ( "bencao", "fu", "ch", "par", ),
    ):
    ivt_name = f"{book}-{bsub}-{utype}-{ltype}"
    target = makef.add_rules_for_units_per_line_file(pre, mak, tit, ivt_name, utype)
    targets.append(target)
  return targets
  # ----------------------------------------------------------------------

def add_all_071_single_hist_rules(pre, mak, tit):
  targets = []
  for book, bsub, utype in ( \
      ( "bencao", "fu", "ch", ),
    ):
    ivt_name = f"{book}-{bsub}-{utype}-par"
    usize = 1.000 # Plot sizes in raw units.
    color = makef.hist_color(book, bsub, utype)
    bin_size = 5 if utype == "ec" else 1 
    target = makef.add_rules_for_single_parag_size_hist_plot(pre, mak, tit, ivt_name, usize, color, bin_size)
    targets.append(target)
  return targets
  # ----------------------------------------------------------------------

def add_all_071_double_hist_rules(pre, mak, tit):
  targets = []
  for book0, bsub0, utype0, book1, bsub1, utype1, raw_size in ( ):
    
    assert utype0 == "ch"
    usize0 = 1
    ivt_name0 = f"{book0}-{bsub0}-{utype0}-par" 
    color0 = makef.hist_color(book0, bsub0, utype0)
    
    assert utype1 == "ch"
    usize1 = 1 
    ivt_name1 = f"{book1}-{bsub1}-{utype1}-par"
    color1 = makef.hist_color(book1, bsub1, utype1)
    
    bin_size = 1 
    target = makef.add_rules_for_double_parag_size_hist_plot \
      ( pre, mak, tit, 
        ivt_name0, usize0, color0, 
        ivt_name1, usize1, color1, 
        bin_size
      )
    targets.append(target)
  return targets
  # ----------------------------------------------------------------------

def add_all_071_bencao_word_pos_plot_rules(pre, mak, tit):
  # Word position plots for various words and various parags in the SBJ.

  book = "bencao"
  bsub = "fu"
  ltype = "par"
  utype = "ch"
  usize = 1.0

  ivt_name = f"{book}-{bsub}-{utype}-{ltype}"
  color = makef.hist_color(book, bsub, utype)

  kword_en_list = ( "USES", )
  
  kword_en_list_not = ( "MAKES", "QI", )
    
  slocs_maybe = ( \
      "b1.4.100", # 3 hits.
      "b2.5.118", # 3 hits?.
      "b2.5.518", # 3 hits?.
      "b3.3.080", # 3 hits.
      "b2.1.013", # 3 hits.
    )
  slocs_yes = ( \
      "b1.4.096", # 7 hits - red rooster.
      "b2.4.094", # 3 hits.
      "b3.3.088", # 3 hits.
    )
  slocs_not = ( \
      "b1.1.014", # 1 hit.
      "b1.2.061", # 1 hit.
      "b1.1.007", # 1 hit.
      "b1.1.014", # 1 hit.
      "b1.2.061", # 1 hit.
      "b1.4.090", # 2 hits.
      "b1.5.106", # 1 hit.
      "b1.5.109", # 2 hits.
      "b2.1.001", # 1 hit.
      "b2.2.066", # 1 hit.
      "b2.3.090", # 2 hits.
      "b3.3.071", # 1 hit.
      "b3.5.119", # 2 hits.
    )

  sloc_list = slocs_maybe + slocs_yes
  
  targets = makef.add_rules_for_many_word_pos_plots \
    ( pre, mak, tit, ivt_name, utype, usize, color, kword_en_list, sloc_list )
  return targets
  # ----------------------------------------------------------------------

def add_all_071_lines_with_pattern_rules(pre, mak, tit):
  # Creates ".opa" files for SBJ recipes that contain certain patterns.

  debug = False
  targets = []
  
  for book, kword_en in ( \
      ('bencao', 'USE1',  ), 
    ):
    assert book == "bencao"
    utype_raw = "ch"; utype_match = "ch"; ltype = "par"; bsub = "fu"
    ivt_name = f"{book}-{bsub}-{utype_raw}-{ltype}"
    target = makef.add_rules_for_lines_with_pattern_files \
      ( pre, mak, tit, ivt_name, utype_match, kword_en )
    if debug: err.write(f"!= {target = }\n")
    targets.append(target)
  return targets
  # ----------------------------------------------------------------------

def add_all_071_word_tuple_rules(pre, mak, tit):
  # Word tuples
  targets = []
  for ivt_name, utype, tsize, kword_en in ( 
      ( "bencao-fu-ch-par", "ch",  1, None,   ),
      ( "bencao-fu-ch-par", "ch",  2, None,   ),
      ( "bencao-fu-ch-par", "ch",  3, None,   ),

      ( "bencao-fu-ch-par", "ch",  4, 'USES', ),
    ):
    target = makef.add_rules_for_word_tuple_file \
      ( pre, mak, tit, ivt_name, utype, tsize, kword_en )
    targets.append(target)
  return targets
  # ----------------------------------------------------------------------