initial commit

.gitignore

@@ -0,0 +1,2 @@
+*.txt
+*.pdf

README.md

@@ -0,0 +1,10 @@
+# ITU Operating Agency Table Parser
+
+The ITU does not appear to publish a machine-readable list of Operating Agencies (except for the dropdown menus in SpaceCap).
+
+To assist automatic ingestion of ITU filing data, this PHP script parses a text dump from the ITU-published PDF file 
+PREFACE_EN.pdf ("PREFACE TO THE BR INTERNATIONAL FREQUENCY INFORMATION CIRCULAR (TERRESTRIAL SERVICES)")
+and generates a properly-formatted CSV of every Operating Agency (and its Administration) discovered.
+
+The script fetch_and_convert.sh will automatically download the latest PREFACE_EN.pdf directly from the ITU website
+(this file is updated regularly), then convert it to a CSV.

dump.php

@@ -0,0 +1,80 @@
+<?php
+// I guarantee I will not understand any of this in 48 hours
+
+$data_dirty = file_get_contents("PREFACE_EN.txt");
+$data_clean = str_replace("\u{000c}", "", $data_dirty);
+file_put_contents("data_clean.txt", $data_clean);
+
+$handle = fopen("data_clean.txt", "r");
+if(!$handle) die("Could not open input file");
+
+$out = fopen('output.csv', 'w');
+if(!$handle) die("Could not open output file");
+
+function attempt_gets(){
+    global $handle, $line;
+
+    if(($fileline = fgets($handle)) !== false){
+        $line = trim($fileline);
+        return true;
+    }else{
+        fclose($handle);
+        die();
+    }
+}
+
+$line = "";
+$agency = "";
+$agency_list = false;
+$in_table = false;
+$adm_finished = false;
+
+while(attempt_gets()) {
+    if($line == "") continue; // ignore blank lines, usually near page markings
+    if($in_table && str_starts_with($line, "SECTION 4")) die(); // we're at the end of the table
+    if(str_starts_with($line, "Chapter ")) continue; // skip page marking lines
+
+    if(str_starts_with($line, "Symbol Geographical Area")){
+        if($agency_list == true){
+            $agency_num = trim(explode("   ", $agency, 2)[0]);
+            $agency_name = trim(explode("   ", $agency, 2)[1]);
+            if(!($cur_adm == "" || $country == "" || $agency_num == "" || $agency_name == "")){
+                //echo("\"$cur_adm\",\"$country\",\"$agency_num\",\"$agency_name\"\n");
+                fputcsv($out, [$cur_adm, $country, $agency_num, $agency_name]);
+            }
+        }
+        $agency_list = false; // we have finished a list of operating agencies, but print the last one
+
+        attempt_gets();
+        $line = trim(preg_split('/ (A|B|C) /', $line)[0]);
+        $cur_adm = trim(explode("     ", $line, 2)[0]);
+        $country = trim(explode("     ", $line, 2)[1]);
+        $adm_finished = true;
+        $in_table = true;
+        //echo("ADMINISTRATION $cur_adm, COUNTRY $country \n");
+    }else if($agency_list && $in_table){
+        if(ctype_digit(substr($line, 0, 3)) && substr($line, 3, 3) == "   "){
+            //echo("Parse Agency $agency\n");
+            //if($agency == "") $agency = $line;
+
+            $agency_num = trim(explode("   ", $agency, 2)[0]);
+            $agency_name = trim(explode("   ", $agency, 2)[1]);
+            //echo("AGENCY " . $agency_num . " NAME " . $agency_name . "\n");
+
+            if(!($cur_adm == "" || $country == "" || $agency_num == "" || $agency_name == "") && !$adm_finished){
+                //echo("\"$cur_adm\",\"$country\",\"$agency_num\",\"$agency_name\"\n");
+                fputcsv($out, [$cur_adm, $country, $agency_num, $agency_name]);
+            }
+            $adm_finished = false;
+
+            $agency = $line;
+        }else{
+            $agency .= " " . $line;
+        }
+    }else if(str_starts_with($line, "12A Code") && $in_table){
+        $agency_list = true; // we are starting a list of operating agencies
+    }
+}
+
+fclose($handle);
+fclose($out);

fetch_and_convert.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+rm PREFACE_EN.pdf
+wget https://www.itu.int/en/ITU-R/terrestrial/brific/BRIFIC/Preface/PREFACE_EN.pdf
+pdftotext -layout PREFACE_EN.pdf
+php dump.php