#! /usr/bin/python3
# Last edited on 2026-10-12 09:37:17 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_072_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_072_double_hist_rules(pre, mak, tit) + \
    add_all_072_coin_image_rules(pre, mak, tit)

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

def add_all_072_double_hist_rules(pre, mak, tit):
  targets = []
  for book0, bsub0, utype0, book1, bsub1, utype1 in ( \
      ("bencao", "fu", "ch", "starps", "fu", "wp", ), # Compare SBJ with SPS "fu-wp".
      ("bencao", "fu", "ch", "starps", "fu", "wc", ), # Compare SBJ with SPS "fu-wc".
      ("bencao", "fu", "ch", "starps", "fu", "ec", ), # Compare SBJ with SPS "fu-ec".
      \
      ("bencao", "fu", "ch", "starps", "gd", "wp", ), # Compare SBJ with SPS "gd-wp".
      ("bencao", "fu", "ch", "starps", "gd", "wc", ), # Compare SBJ with SPS "gd-wc".
      ("bencao", "fu", "ch", "starps", "gd", "ec", ), # Compare SBJ with SPS "gd-ec".
    ):
    
    del_qo = True
    exp_irm = True
    
    ivt_name0 = f"{book0}-{bsub0}-{utype0}-par" 
    usize0 = 1 if utype0 == "ch" else 1/spf.expected_eva_letters_per_hanzi(utype0, del_qo, exp_irm)
    color0 = makef.hist_color(book0, bsub0, utype0)
    
    ivt_name1 = f"{book1}-{bsub1}-{utype1}-par"
    usize1 = 1 if utyp1 == "ch" else 1/spf.expected_eva_letters_per_hanzi(utype1, del_qo, exp_irm)
    color1 = makef.hist_color(book1, bsub1, utype1)
    
    bin_size = 5 if utype0 == "ec" else 1 
    target =  add_072_rules_for_double_parag_size_hist_plot \
      ( pre, mak, tit, 
        ivt_name0, utype0, usize0, color0, 
        ivt_name1, utype1, usize1, color1, 
        bin_size
      )
    targets.append(target)
  return targets
  # ----------------------------------------------------------------------

def add_all_072_coin_image_rules(pre, mak, tit):
  # Adds rules and commands to create files
  # "res/{upp_name0}-{upp_name1}-coin-map.png" that shows coincidences of sizes
  # etc between the sizes of the parags of the SBJ "../071/{upp_name0}-par.upp" and those of the
  # SPS "res/{upp_name1}-par.upp".  Returns the upp_names of the targets (sans "res/").
  
  # The starps file should be first because it is the shortest one.

  utype1 = "ch"
  upp_name1 = f"bencao-fu-{utype1}-par"
  usize1 = 1.000

  targets = []
  for bsub0, utype0 in ("gd", "wc"), ("gd", "wp"), ("gd", "ec"):
    upp_name0 = f"starps-{bsub0}-{utype0}-par"
    del_qo = True
    exp_irm = True
    usize0 = 1/spf.expected_eva_letters_per_hanzi(utype0, del_qo, exp_irm)
    target = add_072_rules_for_parag_parag_coin_image \
      ( pre, mak, tit, 
        upp_name0, utype0, usize0, 
        upp_name1, utype1, usize1
      )
    targets.append(target)
    
  return targets
  # ----------------------------------------------------------------------

def add_072_rules_for_units_per_line_file(pre, mak, tit, ivt_name, utype):

  assert utype == "ch" or utype == "ec" or utype == "wc" or utype == "wp"
  
  # The ".upp" files are created other Notes:
  source_dir = "../071" if utype == "ch" else "../070"
  source = f"{source_dir}/res/{ivt_name}.ivt"

  pfn_py_lib = "work/process_funcs.py"
  cfn_py_lib = "work/chinese_funcs.py"
  
  target = f"{ivt_name}.upp"

  # The size counting script and its imported modules:

  tit[target] = f"making parag size file {source_dir}/res{target} from {source}"
  pre[target] = [ source_dir ]
  mak[target] = (
      f"( cd {source_dir} && make TARGETS='{target}' ) ",
    )
  return target
  # ----------------------------------------------------------------------

def add_072_rules_for_double_parag_size_hist_plot \
  ( pre, mak, tit, 
    ivt_name0, utype0, usize0, color0, 
    ivt_name1, utype1, usize1, color1, 
    bin_size
  ): 
  # Rules to create the plot file "{ivt_name0}-{ivt_name1}-upp-hist.png"
  # that compare the histograms of parag sizes listed in files
  # "{dir0}/res/{ivt_name0}.upp" and "{dir1}/res/es/{ivt_name1}.upp",
  # where {dir0} and {dir1} are "../070" for unit types "ec", "wp", and "wc",
  # "or "../071" for unit type "ch".
  #
  # The unit types {utype0,utype1} must be consistent with {ivt_name0,ivt_name1}

  dir0 = "../071" if utype0 == "ch" else "../070"
  dir1 = "../071" if utype1 == "ch" else "../070"

  # Add rules to create the prerequisite files:
  for ivt_name, utype in (ivt_name0, utype0), (ivt_name1, utype1): 
    pre_target = add_072_rules_for_units_per_line_file(pre, mak, tit, ivt_name, utype)
    assert pre_target == f"{ivt_name}.upp"

  target = f"{ivt_name0}-{ivt_name1}-upp-hist.png"

  # Hstogram plotting script and its imported modules:
  script = "work/plot_double_parag_size_histograms.sh"
  mkwdhist_script = "work/make_hist_of_units_per_parag.sh"
  mkhist_script = "work/make_histogram.gawk"
  h2poly_script = "work/turn_histogram_into_polygonal_line.gawk"
  efn_gawk_lib = "work/error_funcs.gawk"

  pre[target] = [ script, mkwdhist_script, efn_gawk_lib, mkhist_script, h2poly_script ]
  pre[target].append(f"{dir0}/res/{ivt_name0}.upp")
  pre[target].append(f"{dir1}/res/{ivt_name1}.upp")
  tit[target] = f"plotting parag size histograms for {ivt_name0} and {ivt_name1}"
  mak[target] = (
      f"{script} {dir0} {ivt_name0} {usize0:.3f} '{color0}' {dir1} {ivt_name1} {usize1:.3f} '{color1}' {bin_size}",
    )
  return target
  # ----------------------------------------------------------------------

def add_072_rules_for_parag_parag_coin_image \
  ( pre, mak, tit, 
    ivt_name0, utype0, usize0, 
    ivt_name1, utype1, usize1
  ):
  # Adds rules and commands to create a target file
  # "res/{ivt_name0}-{ivt_name1}-coin-map.png" that shows coincidences
  # of sizes between the SPS recipes and SBJ
  # parags.  These sizes are 
  # obtained from
  # "{dir0}/res/{ivt_name0}.upp" and "{dir1}/res/es/{ivt_name1}.upp",
  # where {dir0} and {dir1} are "../070" for unit types "ec", "wp", and "wc",
  # "or "../071" for unit type "ch".

  dir0 = "../071" if utype0 == "ch" else "../070"
  dir1 = "../071" if utype1 == "ch" else "../070"

  # Add rules to create the prerequisite files:
  for ivt_name, utype in (ivt_name0, utype0), (ivt_name1, utype1): 
    pre_target = add_072_rules_for_units_per_line_file(pre, mak, tit, ivt_name, utype)
    assert pre_target == f"{ivt_name}.upp"

  target = f"{ivt_name0}-{ivt_name1}-coin-map.png"

  # Hstogram plotting script and its imported modules:
  script = "create_parag_size_similarity_image.py"

  pre[target] = [ script, ]
  pre[target].append(f"{dir0}/res/{ivt_name0}.upp")
  pre[target].append(f"{dir1}/res/{ivt_name1}.upp")
  tit[target] = f"plotting parag size histograms for {ivt_name0} and {ivt_name1}"
  mak[target] = (
      f"{script} {dir0} {ivt_name0} {utype0} {usize0} {dir1} {ivt_name1} {utype1} {usize1}",
    )
  return target
  # ----------------------------------------------------------------------