Merge pull request #71 from dreamer-coding/main

dreamer-coding · web-flow · commit bd2f1225fcad · 2025-09-07T08:33:19.000-07:00
diff --git a/code/logic/fossil/io/soap.h b/code/logic/fossil/io/soap.h
@@ -136,6 +136,54 @@ int fossil_io_soap_detect_offensive(const char *text);
  */
 int fossil_io_soap_detect_neutral(const char *text);
 
+/**
+ * @brief Detect hype-related phrases in text.
+ *
+ * @param text Input text to scan.
+ * @return Non-zero if hype phrases are detected, 0 otherwise.
+ */
+int fossil_io_soap_detect_hype(const char *text);
+
+/**
+ * @brief Detect quality-related phrases in text.
+ *
+ * @param text Input text to scan.
+ * @return Non-zero if quality phrases are detected, 0 otherwise.
+ */
+int fossil_io_soap_detect_quality(const char *text);
+
+/**
+ * @brief Detect political content in text.
+ *
+ * @param text Input text to scan.
+ * @return Non-zero if political patterns are detected, 0 otherwise.
+ */
+int fossil_io_soap_detect_political(const char *text);
+
+/**
+ * @brief Detect conspiracy-related content in text.
+ *
+ * @param text Input text to scan.
+ * @return Non-zero if conspiracy patterns are detected, 0 otherwise.
+ */
+int fossil_io_soap_detect_conspiracy(const char *text);
+
+/**
+ * @brief Detect marketing/jargon-heavy content in text.
+ *
+ * @param text Input text to scan.
+ * @return Non-zero if marketing patterns are detected, 0 otherwise.
+ */
+int fossil_io_soap_detect_marketing(const char *text);
+
+/**
+ * @brief Detect technobabble or meaningless jargon in text.
+ *
+ * @param text Input text to scan.
+ * @return Non-zero if technobabble patterns are detected, 0 otherwise.
+ */
+int fossil_io_soap_detect_technobabble(const char *text);
+
 // filter functions
 
 /**
@@ -357,6 +405,66 @@ namespace fossil {
                 return fossil_io_soap_detect_neutral(text.c_str()) != 0;
             }
 
+            /**
+             * @brief Detects if a given text contains hype-related content.
+             *
+             * @param text Input string to analyze
+             * @return true if hype detected, false otherwise
+             */
+            static bool is_hype(const std::string &text) {
+                return fossil_io_soap_detect_hype(text.c_str()) != 0;
+            }
+
+            /**
+             * @brief Detects if a given text contains high-quality phrasing.
+             *
+             * @param text Input string to analyze
+             * @return true if quality phrasing detected, false otherwise
+             */
+            static bool is_quality(const std::string &text) {
+                return fossil_io_soap_detect_quality(text.c_str()) != 0;
+            }
+
+            /**
+             * @brief Detects if a given text contains political content.
+             *
+             * @param text Input string to analyze
+             * @return true if political content detected, false otherwise
+             */
+            static bool is_political(const std::string &text) {
+                return fossil_io_soap_detect_political(text.c_str()) != 0;
+            }
+
+            /**
+             * @brief Detects if a given text contains conspiracy-related content.
+             *
+             * @param text Input string to analyze
+             * @return true if conspiracy content detected, false otherwise
+             */
+            static bool is_conspiracy(const std::string &text) {
+                return fossil_io_soap_detect_conspiracy(text.c_str()) != 0;
+            }
+
+            /**
+             * @brief Detects if a given text contains marketing/jargon-heavy content.
+             *
+             * @param text Input string to analyze
+             * @return true if marketing jargon detected, false otherwise
+             */
+            static bool is_marketing(const std::string &text) {
+                return fossil_io_soap_detect_marketing(text.c_str()) != 0;
+            }
+
+            /**
+             * @brief Detects if a given text contains technobabble or meaningless jargon.
+             *
+             * @param text Input string to analyze
+             * @return true if technobabble detected, false otherwise
+             */
+            static bool is_technobabble(const std::string &text) {
+                return fossil_io_soap_detect_technobabble(text.c_str()) != 0;
+            }
+
             /**
              * @brief Fix common grammar errors in input text.
              * 
diff --git a/code/logic/soap.c b/code/logic/soap.c
@@ -373,6 +373,84 @@ static const char *SOAP_NEUTRAL_PATTERNS[] = {
     "clear", "direct", "simple", "uncomplicated", NULL
 };
 
+/** Lookup table for hype phrases */
+static const char *SOAP_HYPE_PATTERNS[] = {
+    "game-changing","revolutionary","cutting-edge","disruptive","never before seen",
+    "unprecedented","the ultimate","next-level","life-changing","epic","legendary",
+    "the best ever","mind-blowing","world-class","exclusive","groundbreaking","innovative",
+    "unparalleled","must-have","hot new","phenomenal","incredible","jaw-dropping",
+    "guaranteed","proven results","miracle","unstoppable","the future of","beyond belief",
+    "state-of-the-art","amazing","once in a lifetime","extraordinary","unbeatable","elite",
+    "remarkable","top-tier","exceptional","stellar","premium","best-in-class","outstanding",
+    "next big thing","ultimate solution","transformative","game-changer","reimagined","breakthrough",
+    "trailblazing","legendary product","redefining","NULL"
+};
+
+/** Political */
+static const char *SOAP_POLITICAL_PATTERNS[] = { 
+    "left-wing","right-wing","liberal agenda","conservative values","fake news media",
+    "mainstream media","deep state","radical","extremist","social justice warrior","tax and spend",
+    "big government","free market","personal freedom","authoritarian","totalitarian","censorship",
+    "political correctness","cancel culture","big brother","elitist","grassroots movement",
+    "identity politics","partisan","special interests","lobbyist","corrupt politicians","vote rigging",
+    "rigged system","polarized nation","propaganda","electoral fraud","media bias","government overreach",
+    "legislation","policy reform","constitutional","impeachment","regulatory capture","voter suppression",
+    "political scandal","public office","government shutdown","state control","national security",
+    "campaign finance","party platform","NULL"
+};
+
+/** Conspiracy */
+static const char *SOAP_CONSPIRACY_PATTERNS[] = {
+    "they don’t want you to know","hidden truth","cover up","shadow government","mind control",
+    "secret society","illuminati","new world order","false flag","plandemic","chemtrails",
+    "crisis actors","fabricated evidence","hollow earth","flat earth","alien cover-up",
+    "government lies","deep underground base","suppressed technology","mass surveillance",
+    "thought police","fabricated by the media","controlled opposition","secret labs",
+    "fake moon landing","reptilian","massive hoax","in on it","they control everything",
+    "behind closed doors","manipulation","hidden agenda","classified files","covert operations",
+    "black ops","unexplained phenomena","shadowy figures","secret programs","disinformation",
+    "conspiratorial","hidden messages","mind games","secret experiments","underground network",
+    "NULL"
+};
+
+/** Marketing */
+static const char *SOAP_MARKETING_PATTERNS[] = {
+    "limited time offer","act now","don’t miss out","guaranteed results","risk free",
+    "sign up today","exclusive deal","free trial","buy one get one","special promotion",
+    "limited stock","save big","lowest price ever","best deal","offer ends soon",
+    "only for today","unlock savings","no obligation","instant access","money back guarantee",
+    "redeem now","hot deal","flash sale","discounted rate","claim your spot","exclusive offer",
+    "limited edition","join thousands","be the first","special savings","new arrival",
+    "get yours now","best value","exclusive bonus","early bird","special launch","today only",
+    "final chance","bonus included","premium package","offer valid","special offer","don’t wait",
+    "hurry up","exclusive access","deal of the day","NULL"
+};
+
+/** Technobabble */
+static const char *SOAP_TECHNOBABBLE_PATTERNS[] = {
+    "synergy","blockchain-enabled","AI-powered","machine learning solution",
+    "next-gen","hyper scalable","paradigm shift","deep tech","seamless integration",
+    "big data","cloud-native","cutting-edge platform","quantum leap","value proposition",
+    "turnkey solution","innovative ecosystem","frictionless","low-hanging fruit",
+    "mission critical","digital transformation","smart disruption","empower users",
+    "end-to-end","future-proof","holistic approach","thought leadership","revolutionize",
+    "strategic alignment","game-changer","intelligent automation","data-driven","disruptive innovation",
+    "scalable architecture","AI-driven","cloud-first","next-generation","hyper convergence",
+    "machine intelligence","tech stack","real-time analytics","dynamic workflow","intelligent design",
+    "agile methodology","NULL"
+};
+
+/** Lookup table for low-quality signals */
+static const char *SOAP_QUALITY_PATTERNS[] = {
+    "just saying","you know","like literally","basically","whatever","stuff and things",
+    "random nonsense","blah blah","and so on","thingy","meh","idk","not sure","somehow",
+    "something like that","kind of","sort of","whatever works","in a way","obviously",
+    "clearly","everyone knows","trust me","believe me","it is what it is","that kind of thing",
+    "doesn’t matter","whatever you think","unimportant","insignificant","minor detail",
+    "whatever floats your boat","trivial","meaningless","small stuff","irrelevant",
+    "empty words","hollow statement","noise","filler","pointless","NULL"
+};
+
 /** Lookup table for words that need to be skipped due to misdetection */
 static const char *SKIP_WORDS[] = {
     "limit", "size", "width", "height", "length", "depth", "volume", "capacity", "weight",
@@ -691,13 +769,19 @@ const char *fossil_io_soap_detect_tone(const char *text) {
         {SOAP_CLICKBAIT_PATTERNS, "clickbait"},
         {SOAP_SPAM_PATTERNS, "spam"},
         {SOAP_WOKE_PATTERNS, "woke"},
-        {SOAP_BOT_PATTERNS, "bot"},
-        {SOAP_SARCASTIC_PATTERNS, "sarcastic"},
-        {SOAP_SNOWFLAKE_PATTERNS, "snowflake"},
-        {SOAP_FORMAL_PATTERNS, "formal"},
-        {SOAP_OFFENSIVE_PATTERNS, "offensive"},
-        {SOAP_NEUTRAL_PATTERNS, "neutral"},
-        {NULL, NULL}
+        {SOAP_BOT_PATTERNS,           "bot"},
+        {SOAP_SARCASTIC_PATTERNS,     "sarcastic"},
+        {SOAP_SNOWFLAKE_PATTERNS,     "snowflake"},
+        {SOAP_FORMAL_PATTERNS,        "formal"},
+        {SOAP_OFFENSIVE_PATTERNS,     "offensive"},
+        {SOAP_NEUTRAL_PATTERNS,       "neutral"},
+        { SOAP_QUALITY_PATTERNS,      "quality"     },
+        { SOAP_HYPE_PATTERNS,         "hype"        },
+        { SOAP_POLITICAL_PATTERNS,    "political"   },
+        { SOAP_CONSPIRACY_PATTERNS,   "conspiracy"  },
+        { SOAP_MARKETING_PATTERNS,    "marketing"   },
+        { SOAP_TECHNOBABBLE_PATTERNS, "technobabble"},
+        { NULL,                       NULL          }
     };
 
     // Split text into words for more accurate matching
@@ -894,6 +978,30 @@ int fossil_io_soap_detect_neutral(const char *text) {
     return soap_detect_patterns(text, SOAP_NEUTRAL_PATTERNS);
 }
 
+int fossil_io_soap_detect_hype(const char *text) {
+    return soap_detect_patterns(text, SOAP_HYPE_PATTERNS);
+}
+
+int fossil_io_soap_detect_quality(const char *text) {
+    return soap_detect_patterns(text, SOAP_QUALITY_PATTERNS);
+}
+
+int fossil_io_soap_detect_political(const char *text) {
+    return soap_detect_patterns(text, SOAP_POLITICAL_PATTERNS);
+}
+
+int fossil_io_soap_detect_conspiracy(const char *text) {
+    return soap_detect_patterns(text, SOAP_CONSPIRACY_PATTERNS);
+}
+
+int fossil_io_soap_detect_marketing(const char *text) {
+    return soap_detect_patterns(text, SOAP_MARKETING_PATTERNS);
+}
+
+int fossil_io_soap_detect_technobabble(const char *text) {
+    return soap_detect_patterns(text, SOAP_TECHNOBABBLE_PATTERNS);
+}
+
 /**
  * @brief Filter text by replacing words/phrases matching any pattern (comma-separated) with '*'.
  *        Patterns support '*' and '?' wildcards, case-insensitive.
diff --git a/code/tests/cases/test_soap.c b/code/tests/cases/test_soap.c
@@ -151,6 +151,84 @@ FOSSIL_TEST(c_test_io_soap_detect_offensive_false) {
     ASSUME_ITS_FALSE(result);
 }
 
+// --- HYPE detection ---
+FOSSIL_TEST(c_test_io_soap_detect_hype_true) {
+    const char *input = "This is the ultimate revolutionary game-changing breakthrough!";
+    int result = fossil_io_soap_detect_hype(input);
+    ASSUME_ITS_TRUE(result);
+}
+
+FOSSIL_TEST(c_test_io_soap_detect_hype_false) {
+    const char *input = "This is a normal system update with minor improvements.";
+    int result = fossil_io_soap_detect_hype(input);
+    ASSUME_ITS_FALSE(result);
+}
+
+// --- QUALITY detection ---
+FOSSIL_TEST(c_test_io_soap_detect_quality_true) {
+    const char *input = "Everyone knows this method is reliable and clearly follows strict methodology.";
+    int result = fossil_io_soap_detect_quality(input);
+    ASSUME_ITS_TRUE(result);
+}
+
+FOSSIL_TEST(c_test_io_soap_detect_quality_false) {
+    const char *input = "Wow, this is amazing and unbelievable!";
+    int result = fossil_io_soap_detect_quality(input);
+    ASSUME_ITS_FALSE(result);
+}
+
+// --- POLITICAL detection ---
+FOSSIL_TEST(c_test_io_soap_detect_political_true) {
+    const char *input = "The government overreach and big government policies affect personal freedom.";
+    int result = fossil_io_soap_detect_political(input);
+    ASSUME_ITS_TRUE(result);
+}
+
+FOSSIL_TEST(c_test_io_soap_detect_political_false) {
+    const char *input = "I enjoy going on long hikes in the mountains and reading books.";
+    int result = fossil_io_soap_detect_political(input);
+    ASSUME_ITS_FALSE(result);
+}
+
+// --- CONSPIRACY detection ---
+FOSSIL_TEST(c_test_io_soap_detect_conspiracy_true) {
+    const char *input = "Hidden truth and secret societies control world events.";
+    int result = fossil_io_soap_detect_conspiracy(input);
+    ASSUME_ITS_TRUE(result);
+}
+
+FOSSIL_TEST(c_test_io_soap_detect_conspiracy_false) {
+    const char *input = "Astronomers study the moon landing and other space phenomena.";
+    int result = fossil_io_soap_detect_conspiracy(input);
+    ASSUME_ITS_FALSE(result);
+}
+
+// --- MARKETING detection ---
+FOSSIL_TEST(c_test_io_soap_detect_marketing_true) {
+    const char *input = "Sign up today for our exclusive limited-time offer!";
+    int result = fossil_io_soap_detect_marketing(input);
+    ASSUME_ITS_TRUE(result);
+}
+
+FOSSIL_TEST(c_test_io_soap_detect_marketing_false) {
+    const char *input = "This is a technical description of a microcontroller circuit.";
+    int result = fossil_io_soap_detect_marketing(input);
+    ASSUME_ITS_FALSE(result);
+}
+
+// --- TECHNOBABBLE detection ---
+FOSSIL_TEST(c_test_io_soap_detect_technobabble_true) {
+    const char *input = "Our cloud-native AI-powered platform enables seamless integration and next-gen innovation.";
+    int result = fossil_io_soap_detect_technobabble(input);
+    ASSUME_ITS_TRUE(result);
+}
+
+FOSSIL_TEST(c_test_io_soap_detect_technobabble_false) {
+    const char *input = "The client connects to the server via a standard HTTPS request.";
+    int result = fossil_io_soap_detect_technobabble(input);
+    ASSUME_ITS_FALSE(result);
+}
+
 // filter cases
 
 FOSSIL_TEST(c_test_io_soap_add_custom_filter) {
@@ -326,6 +404,18 @@ FOSSIL_TEST_GROUP(c_soap_tests) {
     FOSSIL_TEST_ADD(c_soap_suite, c_test_io_soap_detect_snowflake_false);
     FOSSIL_TEST_ADD(c_soap_suite, c_test_io_soap_detect_offensive_true);
     FOSSIL_TEST_ADD(c_soap_suite, c_test_io_soap_detect_offensive_false);
+    FOSSIL_TEST_ADD(c_soap_suite, c_test_io_soap_detect_hype_true);
+    FOSSIL_TEST_ADD(c_soap_suite, c_test_io_soap_detect_hype_false);
+    FOSSIL_TEST_ADD(c_soap_suite, c_test_io_soap_detect_quality_true);
+    FOSSIL_TEST_ADD(c_soap_suite, c_test_io_soap_detect_quality_false);
+    FOSSIL_TEST_ADD(c_soap_suite, c_test_io_soap_detect_political_true);
+    FOSSIL_TEST_ADD(c_soap_suite, c_test_io_soap_detect_political_false);
+    FOSSIL_TEST_ADD(c_soap_suite, c_test_io_soap_detect_conspiracy_true);
+    FOSSIL_TEST_ADD(c_soap_suite, c_test_io_soap_detect_conspiracy_false);
+    FOSSIL_TEST_ADD(c_soap_suite, c_test_io_soap_detect_marketing_true);
+    FOSSIL_TEST_ADD(c_soap_suite, c_test_io_soap_detect_marketing_false);
+    FOSSIL_TEST_ADD(c_soap_suite, c_test_io_soap_detect_technobabble_true);
+    FOSSIL_TEST_ADD(c_soap_suite, c_test_io_soap_detect_technobabble_false);
 
     // filter tests
     FOSSIL_TEST_ADD(c_soap_suite, c_test_io_soap_add_custom_filter);
