move string function out of openAI.cpp into strops.cpp

5 files changed, 103 insertions, 90 deletions

diff --git a/docs/CHANGES.rst b/docs/CHANGES.rst
index 1a46011..b93080c 100644
--- a/docs/CHANGES.rst
+++ b/docs/CHANGES.rst
@@ -13,9 +13,8 @@ for invoice importing using AI:
 - let user choose the model to use in settings/services/ai
 - real error logging for OpenAI and importing in general
 - when importing an invoice: do not accept invoices for unsupported countries (yet)
-
+- write tests for strops.hpp
 - log_set_depth function so data can be grouped
-- move string function out of openAI.cpp into strops.cpp
 - log elapsed time for ai requests
 - refactor _add functions to use _import functions
 - _import functions should not check for validity and should never fail because of invalid data
diff --git a/include/strops.hpp b/include/strops.hpp
index 68a927e..d11f0c2 100644
--- a/include/strops.hpp
+++ b/include/strops.hpp
@@ -24,4 +24,8 @@ void    strops_replace(char *buf, size_t buf_size, const char *search, const cha
 void    strops_replace_int32(char *buf, size_t buf_size, const char *search, int32_t number);
 void    strops_replace_int64(char *buf, size_t buf_size, const char *search, int64_t number);
 void    strops_replace_float(char *buf, size_t buf_size, const char *search, float number, int decimals);
-bool    strops_prefix(const char *pre, const char *str);
\ No newline at end of file
+bool    strops_prefix(const char *pre, const char *str);
+char*   strops_get_json_value(const char *json, const char *key, char *out, size_t out_size, int nth = 0);
+char*   strops_get_filename(const char* path);
+char*   strops_prep_str_for_json(const char *input, size_t buffer_size);
+char*   strops_unprep_str_from_json(char *input);
\ No newline at end of file
diff --git a/src/ai_providers/openAI.cpp b/src/ai_providers/openAI.cpp
index 5dd2c50..e005a80 100644
--- a/src/ai_providers/openAI.cpp
+++ b/src/ai_providers/openAI.cpp
@@ -14,97 +14,21 @@
 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 */
 
+#define _CRT_SECURE_NO_WARNINGS
+
 #include <fstream>
 #include <iostream>
 #include <string>
 
 #define CPPHTTPLIB_OPENSSL_SUPPORT
 #include "httplib.h"
+#include "strops.hpp"
 #include "log.hpp"
 #include "ai_service.hpp"
 
-static char *extract_json_value(const char *json, const char *key, char *out, size_t out_size, int skip = 0) {
-    char pattern[128];
-    snprintf(pattern, sizeof(pattern), "\"%s\"", key);
-    const char *pos = strstr(json, pattern);
-    while(skip > 0) {
-        pos = strstr(pos+1, pattern);
-        skip--; 
-    }
-    if (!pos) return NULL;
-    pos = strchr(pos, ':');
-    if (!pos) return NULL;
-    pos++;
-
-    // Skip whitespace and quotes
-    while (*pos == ' ' || *pos == '\"') pos++;
-
-    size_t i = 0;
-    while (*pos && !(*pos == '\"' && *(pos-1) != '\\') && i < out_size - 1) {
-        out[i++] = *pos++;
-    }
-    out[i] = '\0';
-    return out;
-}
-
-static const char* get_filename(const char* path) {
-    const char* filename = strrchr(path, '/'); // for Unix-style paths
-    if (filename) return filename + 1;         // skip the '/'
-    filename = strrchr(path, '\\');           // for Windows-style paths
-    if (filename) return filename + 1;
-    return path;                               // no slashes found, path itself is filename
-}
-
-static char *escape_quotes(const char *input, size_t buffer_size) {
-    if (!input) return NULL;
-
-    char *result = (char*)malloc(buffer_size + 100); // Ballpark
-    if (!result) return NULL;
-
-    const char *src = input;
-    char *dst = result;
-
-    while (*src) {
-        if (*src == '"') {
-            *dst++ = '\\';
-            *dst++ = '"';
-        } 
-        else if (*src == '\n') {
-            // empty
-        }
-        else {
-            *dst++ = *src;
-        }
-        src++;
-    }
-    *dst = '\0';
-
-    return result;
-}
-
-static char *unescape_quotes(char *input) {
-    if (!input) return NULL;
-
-    char *src = input;
-    char *dst = input;
-
-    while (*src) {
-        if (*src == '\\' && *(src+1) == '"') {
-            src++;
-        }
-        else if (*src == '\\' && *(src+1) == 'n') {
-            src++;src++;
-        } 
-        *dst++ = *src++;
-    }
-    *dst = '\0';
-
-    return input;
-}
-
 static bool _openAI_query_with_file(char* query, size_t query_length, char* file_id, char** response)
 {
-    #define TESTING_IMPORT
+    //#define TESTING_IMPORT
 
     #ifndef TESTING_IMPORT
     const char *api_key = administration_get_ai_service().api_key_public;
@@ -112,7 +36,7 @@ static bool _openAI_query_with_file(char* query, size_t query_length, char* file
     httplib::SSLClient cli("api.openai.com", 443);
     //cli.enable_server_certificate_verification(false);
 
-    char* query_escaped = escape_quotes(query, query_length);
+    char* query_escaped = strops_prep_str_for_json(query, query_length);
     free(query);
     
     size_t body_size = query_length + 200;
@@ -141,8 +65,8 @@ static bool _openAI_query_with_file(char* query, size_t query_length, char* file
     memset(*response, 0, 100000);
     strncpy(*response, response_body, 100000);
     
-    extract_json_value(*response, "text", *response, 100000);
-    *response = unescape_quotes(*response);
+    strops_get_json_value(*response, "text", *response, 100000);
+    *response = strops_unprep_str_from_json(*response);
 
     return 1;
 }
@@ -150,7 +74,7 @@ static bool _openAI_query_with_file(char* query, size_t query_length, char* file
 static bool _openAI_upload_file(char* file_path, char* file_id, size_t file_id_len)
 {
     const char *api_key = administration_get_ai_service().api_key_public;
-    const char *filename = get_filename(file_path);
+    const char *filename = strops_get_filename(file_path);
 
     FILE* orig_file = fopen(file_path, "rb");
     if (orig_file == NULL) {
@@ -179,7 +103,7 @@ static bool _openAI_upload_file(char* file_path, char* file_id, size_t file_id_l
     }
 
     char upload_id[128];
-    extract_json_value(res->body.c_str(), "id", upload_id, sizeof(upload_id));
+    strops_get_json_value(res->body.c_str(), "id", upload_id, sizeof(upload_id));
     size_t part_size = 64000000; // 64mb
     log_info("Created upload %s with part size %zu.", upload_id, part_size);
 
@@ -215,7 +139,7 @@ static bool _openAI_upload_file(char* file_path, char* file_id, size_t file_id_l
         }
         else {
             char part_id[128];
-            extract_json_value(part_res->body.c_str(), "id", part_id, sizeof(part_id));
+            strops_get_json_value(part_res->body.c_str(), "id", part_id, sizeof(part_id));
             if (part_number == 0) snprintf(completion_body+strlen(completion_body), sizeof(completion_body)-strlen(completion_body), "\"%s\"", part_id);
             if (part_number != 0) snprintf(completion_body+strlen(completion_body), sizeof(completion_body)-strlen(completion_body), ", \"%s\"", part_id);
         }
@@ -241,7 +165,7 @@ static bool _openAI_upload_file(char* file_path, char* file_id, size_t file_id_l
     }
 
     char* completion_body_response = (char*)complete_res->body.c_str();
-    extract_json_value(completion_body_response, "id", file_id, file_id_len, 1);
+    strops_get_json_value(completion_body_response, "id", file_id, file_id_len, 1);
 
     return 1;
 }
diff --git a/src/ai_service.cpp b/src/ai_service.cpp
index e271fcb..82d81d6 100644
--- a/src/ai_service.cpp
+++ b/src/ai_service.cpp
@@ -14,6 +14,8 @@
 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 */
 
+#define _CRT_SECURE_NO_WARNINGS
+
 #include <fstream>
 #include <iostream>
 #include <string>
diff --git a/src/strops.cpp b/src/strops.cpp
index 2c5f0b9..28f9bcc 100644
--- a/src/strops.cpp
+++ b/src/strops.cpp
@@ -14,6 +14,7 @@
 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 */
 
+#include <stdlib.h>
 #include <string.h>
 #include <ctype.h>
 #include <stdio.h>
@@ -107,4 +108,87 @@ void strops_replace_float(char *buf, size_t buf_size, const char *search, float
     char num_buf[200];
     snprintf(num_buf, 200, "%.*f", decimals, number);
     strops_replace(buf, buf_size, search, num_buf);
+}
+
+char* strops_get_json_value(const char *json, const char *key, char *out, size_t out_size, int skip)
+{
+    char pattern[128];
+    snprintf(pattern, sizeof(pattern), "\"%s\"", key);
+    const char *pos = strstr(json, pattern);
+    while(skip > 0) {
+        pos = strstr(pos+1, pattern);
+        skip--; 
+    }
+    if (!pos) return NULL;
+    pos = strchr(pos, ':');
+    if (!pos) return NULL;
+    pos++;
+
+    // Skip whitespace and quotes
+    while (*pos == ' ' || *pos == '\"') pos++;
+
+    size_t i = 0;
+    while (*pos && !(*pos == '\"' && *(pos-1) != '\\') && i < out_size - 1) {
+        out[i++] = *pos++;
+    }
+    out[i] = '\0';
+    return out;
+}
+
+char* strops_get_filename(const char* path)
+{
+    char* filename = (char*)strrchr(path, '/'); // for Unix-style paths
+    if (filename) return filename + 1;         // skip the '/'
+    filename = (char*)strrchr(path, '\\');     // for Windows-style paths
+    if (filename) return filename + 1;
+    return (char*)path;                               // no slashes found, path itself is filename
+}
+
+char* strops_prep_str_for_json(const char *input, size_t buffer_size)
+{
+    if (!input) return NULL;
+
+    char *result = (char*)malloc(buffer_size * 2 + 1);
+    if (!result) return NULL;
+
+    const char *src = input;
+    char *dst = result;
+
+    while (*src) {
+        if (*src == '"') {
+            *dst++ = '\\';
+            *dst++ = '"';
+        } 
+        else if (*src == '\n') {
+            // empty
+        }
+        else {
+            *dst++ = *src;
+        }
+        src++;
+    }
+    *dst = '\0';
+
+    return result;
+}
+
+char* strops_unprep_str_from_json(char *input)
+{
+    if (!input) return NULL;
+
+    char *src = input;
+    char *dst = input;
+
+    while (*src) {
+        if (*src == '\\' && *(src+1) == '"') {
+            src++;
+        }
+        else if (*src == '\\' && *(src+1) == 'n') {
+            src++;src++;
+        } 
+        *dst++ = *src++;
+    }
+    *dst = '\0';
+
+    return input;
 }
\ No newline at end of file