From 560db14a85c43d91eb6254ef588eea74bddacedf Mon Sep 17 00:00:00 2001
From: Mitja HORVAT <pinkfluid@gmail.com>
Date: Sun, 14 Nov 2021 09:12:20 +0100
Subject: [PATCH] c: Finish the phogen module

This commit implements the `phogen_encode` function, which takes a bhash
as argument and generates a phonetic string.
---
 c/phogen/inc/phogen.h            |  6 +++
 c/phogen/meson.build             |  3 +-
 c/phogen/phogen_map/phogen_map.c |  1 -
 c/phogen/src/phogen.c            | 72 +++++++++++++++++++++++++++++++-
 4 files changed, 78 insertions(+), 4 deletions(-)

diff --git a/c/phogen/inc/phogen.h b/c/phogen/inc/phogen.h
index 29b27bc..67ccf2c 100644
--- a/c/phogen/inc/phogen.h
+++ b/c/phogen/inc/phogen.h
@@ -1,4 +1,10 @@
 #ifndef PHOGEN_H_INCLUDED
 #define PHOGEN_H_INCLUDED
 
+#include <stdbool.h>
+
+#include "bhash.h"
+
+bool phogen_encode(char *out, size_t out_len, bhash_t *bh_in);
+
 #endif /* PHOGEN_H_INCLUDED */
diff --git a/c/phogen/meson.build b/c/phogen/meson.build
index 393d9fd..e8d24f8 100644
--- a/c/phogen/meson.build
+++ b/c/phogen/meson.build
@@ -13,6 +13,7 @@ phogen_inc = include_directories('inc')
 phogen_lib = static_library(
         'phogen',
         [phogen_map_h, 'src/phogen.c'],
-        include_directories : phogen_inc)
+        include_directories : phogen_inc,
+        dependencies: bhash_dep)
 
 phogen_dep = declare_dependency(link_with : phogen_lib, include_directories : phogen_inc)
diff --git a/c/phogen/phogen_map/phogen_map.c b/c/phogen/phogen_map/phogen_map.c
index 6b607e3..db06558 100644
--- a/c/phogen/phogen_map/phogen_map.c
+++ b/c/phogen/phogen_map/phogen_map.c
@@ -452,7 +452,6 @@ void phogen_rstrip(char *str, char *what)
     }
 }
 
-
 /*
  * Take the buffer in `bigin` and treat is as a big-endian big number.
  * Perform a division using the 32-bit divisor in `base` and return the
diff --git a/c/phogen/src/phogen.c b/c/phogen/src/phogen.c
index e25eb8a..fb8b529 100644
--- a/c/phogen/src/phogen.c
+++ b/c/phogen/src/phogen.c
@@ -1,10 +1,78 @@
-#include <stdlib.h>
-
 /*
+ * =========================================================================
  * Phonetic generator -- used to convert binary data to semi-pronouncable
  * strings
+ * =========================================================================
  */
+#include <stdlib.h>
+#include <stdio.h>
 
 #include "phogen_map.h"
 #include "phogen.h"
 
+/* Number of letters in the english alphabet */
+#define PHOGEN_ENGLISH_NUM  ('z' - 'a' + 1)
+/* Number of valid characters in a n gram */
+#define PHOGEN_NGRAM_CHARS (PHOGEN_ENGLISH_NUM + 1)
+
+static int phogen_ngram_to_index(const char *word);
+
+/*
+ * Create a phonetic string of size `out_sz` by consuming bits from the bhash
+ * `bh_in` and write it to out.
+ */
+bool phogen_encode(char *out, size_t out_sz, bhash_t *bh_in)
+{
+    size_t ii;
+    size_t nmap;
+    uint32_t nm;
+
+    char ngram[3] = "  ";
+
+    if (out_sz == 0) return true;
+
+    for (ii = 0; ii < (out_sz - 1); ii++)
+    {
+        int ni = phogen_ngram_to_index(ngram);
+
+        /* Calculte the number of letters in the current map */
+        for (nmap = 0; nmap < PHOGEN_ENGLISH_NUM; nmap++)
+        {
+            if (phogen_map[ni].map[nmap] == '\0') break;
+        }
+
+        /* Get the index of the next phonetic letter */
+        nm = bhash_mod32(bh_in, nmap);
+        if (nm == BHASH_MOD32_ERR)
+        {
+            fprintf(stderr, "bhash error, unable to compute next phoentic char.");
+            return false;
+        }
+
+        out[ii] = phogen_map[ni].map[nm];
+
+        /* Calculate the next ngram */
+        ngram[0] = ngram[1];
+        ngram[1] = out[ii];
+    }
+
+    out[ii] = '\0';
+
+    return true;
+}
+
+/*
+ * The phogen_map table is fixed size and sorted. We can calculate the offset
+ * of a ngram with the simple calculation below.
+ */
+int phogen_ngram_to_index(const char *word)
+{
+    int i;
+
+    i = (word[0] == ' ' ? 0 : word[0] - 'a' + 1);
+    i *= PHOGEN_NGRAM_CHARS;
+    i += (word[1] == ' ' ? 0 : word[1] - 'a' + 1);
+
+    return i;
+}
+