From 8881649aca9d5d7efca9730dc8ed1098cd3c69d5 Mon Sep 17 00:00:00 2001
From: ado <adnan.abd0124@gmail.com>
Date: Fri, 23 Feb 2024 23:25:03 +0100
Subject: [PATCH] Update get_line_buffer, update new version of get_line_file
 to work with data that has no new line at eof

---
 include/ss/common.hpp   |  9 ++++---
 include/ss/parser.hpp   | 54 +++++++++++++++++++++--------------------
 test/test_parser1_1.cpp | 13 ++++++++++
 3 files changed, 46 insertions(+), 30 deletions(-)

diff --git a/include/ss/common.hpp b/include/ss/common.hpp
index 95953c6..4493cc3 100644
--- a/include/ss/common.hpp
+++ b/include/ss/common.hpp
@@ -58,14 +58,15 @@ ssize_t get_line_file(char** lineptr, size_t* n, FILE* fp) {
 
     (*lineptr)[0] = '\0';
 
+    size_t line_used = 0;
     while (fgets(buff, sizeof(buff), fp) != nullptr) {
-        size_t line_used = strlen(*lineptr);
+        line_used = strlen(*lineptr);
         size_t buff_used = strlen(buff);
 
-        if (*n < buff_used + line_used) {
+        if (*n <= buff_used + line_used) {
             size_t new_n = *n * 2;
 
-            auto new_lineptr = static_cast<char*>(realloc(*lineptr, *n));
+            auto new_lineptr = static_cast<char*>(realloc(*lineptr, new_n));
             if (new_lineptr == nullptr) {
                 errno = ENOMEM;
                 return -1;
@@ -84,7 +85,7 @@ ssize_t get_line_file(char** lineptr, size_t* n, FILE* fp) {
         }
     }
 
-    return -1;
+    return (line_used != 0) ? line_used : -1;
 }
 
 #endif
diff --git a/include/ss/parser.hpp b/include/ss/parser.hpp
index e11969e..c19b834 100644
--- a/include/ss/parser.hpp
+++ b/include/ss/parser.hpp
@@ -744,50 +744,52 @@ private:
         reader& operator=(const reader& other) = delete;
 
         ssize_t get_line_buffer(char** lineptr, size_t* n,
-                                const char* const buffer, size_t csv_data_size,
-                                size_t& curr_char) {
-            size_t pos;
-            int c;
+                                const char* const csv_data_buffer,
+                                size_t csv_data_size, size_t& curr_char) {
+            if (lineptr == nullptr || n == nullptr ||
+                csv_data_buffer == nullptr) {
+                errno = EINVAL;
+                return -1;
+            }
 
             if (curr_char >= csv_data_size) {
                 return -1;
             }
-            c = buffer[curr_char++];
 
-            if (*lineptr == nullptr) {
-                *lineptr =
-                    static_cast<char*>(malloc(get_line_initial_buffer_size));
-                if (*lineptr == nullptr) {
+            if (*lineptr == nullptr || *n < get_line_initial_buffer_size) {
+                auto new_lineptr = static_cast<char*>(
+                    realloc(*lineptr, get_line_initial_buffer_size));
+                if (new_lineptr == nullptr) {
                     return -1;
                 }
-                *n = 128;
+                *lineptr = new_lineptr;
+                *n = get_line_initial_buffer_size;
             }
 
-            pos = 0;
+            size_t line_used = 0;
             while (curr_char <= csv_data_size) {
-                if (pos + 1 >= *n) {
-                    size_t new_size = *n + (*n >> 2);
-                    if (new_size < get_line_initial_buffer_size) {
-                        new_size = get_line_initial_buffer_size;
-                    }
-                    char* new_ptr = static_cast<char*>(
-                        realloc(static_cast<void*>(*lineptr), new_size));
-                    if (new_ptr == nullptr) {
+                if (line_used + 1 >= *n) {
+                    size_t new_n = *n * 2;
+
+                    char* new_lineptr =
+                        static_cast<char*>(realloc(*lineptr, new_n));
+                    if (new_lineptr == nullptr) {
+                        errno = ENOMEM;
                         return -1;
                     }
-                    *n = new_size;
-                    *lineptr = new_ptr;
+                    *n = new_n;
+                    *lineptr = new_lineptr;
                 }
 
-                (*lineptr)[pos++] = c;
+                auto c = csv_data_buffer[curr_char++];
+                (*lineptr)[line_used++] = c;
                 if (c == '\n') {
-                    break;
+                    (*lineptr)[line_used] = '\0';
+                    return line_used;
                 }
-                c = buffer[curr_char++];
             }
 
-            (*lineptr)[pos] = '\0';
-            return pos;
+            return (line_used != 0) ? line_used : -1;
         }
 
         // read next line each time in order to set eof_
diff --git a/test/test_parser1_1.cpp b/test/test_parser1_1.cpp
index 4c8d476..47adbd2 100644
--- a/test/test_parser1_1.cpp
+++ b/test/test_parser1_1.cpp
@@ -571,9 +571,22 @@ template <bool buffer_mode>
 void test_no_new_line_at_eof() {
     test_no_new_line_at_eof_impl<buffer_mode>({});
     test_no_new_line_at_eof_impl<buffer_mode>({{1, 2, "X"}});
+    test_no_new_line_at_eof_impl<buffer_mode>({{1, 2, "X"}, {}});
     test_no_new_line_at_eof_impl<buffer_mode>({{1, 2, "X"}, {3, 4, "YY"}});
+    test_no_new_line_at_eof_impl<buffer_mode>({{1, 2, "X"}, {3, 4, "YY"}, {}});
     test_no_new_line_at_eof_impl<buffer_mode>(
         {{1, 2, "X"}, {3, 4, "YY"}, {5, 6, "ZZZ"}, {7, 8, "UUU"}});
+
+    for (size_t i = 0; i < 2 * ss::get_line_initial_buffer_size; ++i) {
+        test_no_new_line_at_eof_impl<buffer_mode>(
+            {{1, 2, std::string(i, 'X')}});
+
+        for (size_t j = 0; j < 2 * ss::get_line_initial_buffer_size; ++j) {
+
+            test_no_new_line_at_eof_impl<buffer_mode>(
+                {{1, 2, std::string(i, 'X')}, {3, 4, std::string(j, 'Y')}});
+        }
+    }
 }
 
 TEST_CASE("test no new line at end of data") {