nv-ingest-api 2025.8.14.dev20250814__py3-none-any.whl → 2025.8.16.dev20250816__py3-none-any.whl

nv_ingest_api/util/string_processing/yaml.py

@@ -0,0 +1,45 @@
+import os
+import re
+
+# This regex finds all forms of environment variables:
+# $VAR, ${VAR}, $VAR|default, and ${VAR|default}
+# It avoids matching escaped variables like $$.
+# Default values can be quoted or unquoted.
+_ENV_VAR_PATTERN = re.compile(
+    r"""(?<!\$)\$(?:
+        {(?P<braced>\w+)(?:\|(?P<braced_default>[^}]+))?}
+        |
+        (?P<named>\w+)(?:\|(?P<named_default>"[^"\\]*(?:\\.[^"\\]*)*"|'[^'\\]*(?:\\.[^'\\]*)*'|\S+))?
+    )""",
+    re.VERBOSE,
+)
+
+
+def _replacer(match: re.Match) -> str:
+    """Replaces a regex match with the corresponding environment variable."""
+    var_name = match.group("braced") or match.group("named")
+    default_val = match.group("braced_default") or match.group("named_default")
+
+    # Get value from environment, or use default.
+    value = os.environ.get(var_name, default_val)
+
+    if value is None:
+        return ""
+    return value
+
+
+def substitute_env_vars_in_yaml_content(raw_content: str) -> str:
+    """
+    Substitutes environment variables in a YAML string.
+
+    This function finds all occurrences of environment variable placeholders
+    ($VAR, ${VAR}, $VAR|default, ${VAR|default}) in the input string
+    and replaces them with their corresponding environment variable values.
+
+    Args:
+        raw_content: The raw string content of a YAML file.
+
+    Returns:
+        The YAML string with environment variables substituted.
+    """
+    return _ENV_VAR_PATTERN.sub(_replacer, raw_content)

nv_ingest_api/util/system/hardware_info.py

@@ -16,6 +16,11 @@ CGROUP_V1_CPU_DIR = "/sys/fs/cgroup/cpu"
 CGROUP_V1_CPUACCT_DIR = "/sys/fs/cgroup/cpuacct"  # Sometimes usage is here
 CGROUP_V2_CPU_FILE = "/sys/fs/cgroup/cpu.max"  # Standard path in v2 unified hierarchy
 
+# Memory cgroup paths
+CGROUP_V1_MEMORY_DIR = "/sys/fs/cgroup/memory"
+CGROUP_V2_MEMORY_FILE = "/sys/fs/cgroup/memory.max"  # v2 unified hierarchy
+CGROUP_V2_MEMORY_CURRENT = "/sys/fs/cgroup/memory.current"  # Current usage in v2
+
 
 class SystemResourceProbe:
     """
@@ -70,6 +75,13 @@ class SystemResourceProbe:
         self.cgroup_usage_percpu_us: Optional[list[int]] = None
         self.cgroup_usage_total_us: Optional[int] = None
 
+        # Memory Info
+        self.os_total_memory_bytes: Optional[int] = None
+        self.cgroup_memory_limit_bytes: Optional[int] = None
+        self.cgroup_memory_usage_bytes: Optional[int] = None
+        self.effective_memory_bytes: Optional[int] = None
+        self.memory_detection_method: str = "unknown"
+
         # --- Result ---
         # Raw limit before potential weighting
         self.raw_limit_value: Optional[float] = None
@@ -134,13 +146,13 @@ class SystemResourceProbe:
 
             if quota_us > 0 and period_us > 0:
                 self.cgroup_quota_cores = quota_us / period_us
-                logger.info(
+                logger.debug(
                     f"Cgroup v1 quota detected: {quota_us} us / {period_us} us = {self.cgroup_quota_cores:.2f}"
                     f" effective cores"
                 )
                 return True
             elif quota_us == -1:
-                logger.info("Cgroup v1 quota detected: Unlimited (-1)")
+                logger.debug("Cgroup v1 quota detected: Unlimited (-1)")
                 # No quota limit, but we know it's cgroup v1
                 return True  # Return true because we identified the type
             else:
@@ -149,7 +161,7 @@ class SystemResourceProbe:
         elif shares is not None:  # If only shares are readable, still note it's v1
             self.cgroup_type = "v1"
             self.cgroup_shares = shares
-            logger.info(f"Cgroup v1 shares detected: {shares} (no quota found)")
+            logger.debug(f"Cgroup v1 shares detected: {shares} (no quota found)")
             return True
 
         return False
@@ -171,13 +183,13 @@ class SystemResourceProbe:
                     period_us = int(period_str)
                     self.cgroup_period_us = period_us
                     if quota_str == "max":
-                        logger.info("Cgroup v2 quota detected: Unlimited ('max')")
+                        logger.debug("Cgroup v2 quota detected: Unlimited ('max')")
                         return True  # Identified type, no quota limit
                     else:
                         quota_us = int(quota_str)
                         if quota_us > 0 and period_us > 0:
                             self.cgroup_quota_cores = quota_us / period_us
-                            logger.info(
+                            logger.debug(
                                 f"Cgroup v2 quota detected: {quota_us} us / {period_us}"
                                 f" us = {self.cgroup_quota_cores:.2f} effective cores"
                             )
@@ -204,7 +216,7 @@ class SystemResourceProbe:
             affinity = os.sched_getaffinity(0)  # 0 for current process
             count = len(affinity)
             if count > 0:
-                logger.info(f"Detected {count} cores via os.sched_getaffinity.")
+                logger.debug(f"Detected {count} cores via os.sched_getaffinity.")
                 return count
             else:
                 logger.warning("os.sched_getaffinity(0) returned 0 or empty set.")
@@ -247,9 +259,9 @@ class SystemResourceProbe:
                 logger.error(f"os.cpu_count() failed: {e}")
 
         if logical:
-            logger.info(f"Detected {logical} logical cores via {source}.")
+            logger.debug(f"Detected {logical} logical cores via {source}.")
         if physical:
-            logger.info(f"Detected {physical} physical cores via {source}.")
+            logger.debug(f"Detected {physical} physical cores via {source}.")
 
         return logical, physical
 
@@ -282,7 +294,7 @@ class SystemResourceProbe:
             weighted_cores = (physical_part * 1.0) + (hyperthread_part * self.hyperthread_weight)
 
             if weighted_cores != N:  # Log only if weighting changes the value
-                logger.info(
+                logger.debug(
                     f"Applying hyperthread weight ({self.hyperthread_weight:.2f}) to "
                     f"logical limit {logical_limit} (System: {P}P/{self.os_logical_cores}L): "
                     f"Effective weighted cores = {weighted_cores:.2f}"
@@ -308,6 +320,137 @@ class SystemResourceProbe:
             logger.debug(f"Skipping hyperthread weight calculation for logical limit {logical_limit}.")
             return float(logical_limit)  # Return the original limit as float
 
+    # --- Memory Detection Methods ---
+    @staticmethod
+    def _get_os_memory() -> Optional[int]:
+        """Gets total system memory in bytes using psutil or /proc/meminfo."""
+        # Try psutil first
+        if psutil:
+            try:
+                memory = psutil.virtual_memory()
+                total_bytes = memory.total
+                if total_bytes and total_bytes > 0:
+                    logger.debug(f"Detected {total_bytes / (1024**3):.2f} GB system memory via psutil.")
+                    return total_bytes
+            except Exception as e:
+                logger.warning(f"psutil.virtual_memory() failed: {e}. Falling back to /proc/meminfo.")
+
+        # Fallback to /proc/meminfo
+        try:
+            if os.path.exists("/proc/meminfo"):
+                with open("/proc/meminfo", "r") as f:
+                    for line in f:
+                        if line.startswith("MemTotal:"):
+                            # MemTotal is in KB
+                            parts = line.split()
+                            if len(parts) >= 2:
+                                total_kb = int(parts[1])
+                                total_bytes = total_kb * 1024
+                                logger.debug(
+                                    f"Detected {total_bytes / (1024**3):.2f} GB system memory via /proc/meminfo."
+                                )
+                                return total_bytes
+                            break
+        except (IOError, ValueError, PermissionError) as e:
+            logger.warning(f"Failed to read /proc/meminfo: {e}")
+
+        logger.error("Could not determine system memory from any source.")
+        return None
+
+    def _read_memory_cgroup_v2(self) -> bool:
+        """Attempts to read Cgroup v2 memory limits."""
+        if not os.path.exists(CGROUP_V2_MEMORY_FILE):
+            logger.debug(f"Cgroup v2 memory.max file not found: {CGROUP_V2_MEMORY_FILE}")
+            return False
+
+        logger.debug(f"Checking Cgroup v2 memory limits in {CGROUP_V2_MEMORY_FILE}")
+        content = self._read_file_str(CGROUP_V2_MEMORY_FILE)
+        if content:
+            try:
+                if content == "max":
+                    logger.debug("Cgroup v2 memory limit: unlimited")
+                    return True
+                else:
+                    limit_bytes = int(content)
+                    self.cgroup_memory_limit_bytes = limit_bytes
+                    logger.debug(f"Cgroup v2 memory limit: {limit_bytes / (1024**3):.2f} GB")
+
+                    # Also try to read current usage
+                    usage_content = self._read_file_str(CGROUP_V2_MEMORY_CURRENT)
+                    if usage_content:
+                        try:
+                            usage_bytes = int(usage_content)
+                            self.cgroup_memory_usage_bytes = usage_bytes
+                            logger.debug(f"Cgroup v2 memory usage: {usage_bytes / (1024**3):.2f} GB")
+                        except ValueError:
+                            logger.debug(f"Could not parse memory.current: '{usage_content}'")
+
+                    return True
+            except ValueError:
+                logger.warning(f"Could not parse Cgroup v2 memory.max content: '{content}'")
+        return False
+
+    def _read_memory_cgroup_v1(self) -> bool:
+        """Attempts to read Cgroup v1 memory limits."""
+        if not os.path.exists(CGROUP_V1_MEMORY_DIR):
+            logger.debug(f"Cgroup v1 memory dir not found: {CGROUP_V1_MEMORY_DIR}")
+            return False
+
+        logger.debug(f"Checking Cgroup v1 memory limits in {CGROUP_V1_MEMORY_DIR}")
+
+        # Try memory.limit_in_bytes
+        limit_bytes = self._read_file_int(os.path.join(CGROUP_V1_MEMORY_DIR, "memory.limit_in_bytes"))
+        usage_bytes = self._read_file_int(os.path.join(CGROUP_V1_MEMORY_DIR, "memory.usage_in_bytes"))
+
+        if limit_bytes is not None:
+            # Cgroup v1 often shows very large values (like 9223372036854775807) for unlimited
+            # We consider values >= 2^63-1 or >= system memory * 100 as unlimited
+            if limit_bytes >= 9223372036854775807 or (
+                self.os_total_memory_bytes and limit_bytes >= self.os_total_memory_bytes * 100
+            ):
+                logger.debug("Cgroup v1 memory limit: unlimited (very large value)")
+                return True
+            else:
+                self.cgroup_memory_limit_bytes = limit_bytes
+                logger.debug(f"Cgroup v1 memory limit: {limit_bytes / (1024**3):.2f} GB")
+
+                if usage_bytes is not None:
+                    self.cgroup_memory_usage_bytes = usage_bytes
+                    logger.debug(f"Cgroup v1 memory usage: {usage_bytes / (1024**3):.2f} GB")
+
+                return True
+
+        return False
+
+    def _detect_memory(self):
+        """Performs memory detection sequence."""
+        logger.debug("Starting memory detection...")
+
+        # 1. Get OS level memory first
+        self.os_total_memory_bytes = self._get_os_memory()
+
+        # 2. Try Cgroup v2 memory limits
+        cgroup_memory_detected = self._read_memory_cgroup_v2()
+
+        # 3. Try Cgroup v1 if v2 not found or didn't yield a limit
+        if not cgroup_memory_detected or self.cgroup_memory_limit_bytes is None:
+            cgroup_memory_detected = self._read_memory_cgroup_v1()
+
+        # 4. Determine effective memory
+        if self.cgroup_memory_limit_bytes is not None and self.os_total_memory_bytes is not None:
+            # Use the smaller of cgroup limit and system memory
+            self.effective_memory_bytes = min(self.cgroup_memory_limit_bytes, self.os_total_memory_bytes)
+            self.memory_detection_method = "cgroup_limited"
+            logger.debug(f"Effective memory: {self.effective_memory_bytes / (1024**3):.2f} GB (cgroup limited)")
+        elif self.os_total_memory_bytes is not None:
+            # No cgroup limit, use system memory
+            self.effective_memory_bytes = self.os_total_memory_bytes
+            self.memory_detection_method = "system_memory"
+            logger.debug(f"Effective memory: {self.effective_memory_bytes / (1024**3):.2f} GB (system memory)")
+        else:
+            logger.error("Could not determine effective memory limit")
+            self.memory_detection_method = "failed"
+
     def _detect(self):
         """Performs the detection sequence and applies weighting."""
         logger.debug("Starting effective core count detection...")
@@ -325,7 +468,10 @@ class SystemResourceProbe:
         # 4. Get OS Affinity
         self.os_sched_affinity_cores = self._get_os_affinity()
 
-        # --- 5. Determine the RAW Limit (before weighting) ---
+        # 5. Detect Memory
+        self._detect_memory()
+
+        # --- 6. Determine the RAW Limit (before weighting) ---
         raw_limit = float("inf")
         raw_method = "unknown"
 
@@ -361,9 +507,9 @@ class SystemResourceProbe:
 
         self.raw_limit_value = raw_limit
         self.raw_limit_method = raw_method
-        logger.info(f"Raw CPU limit determined: {self.raw_limit_value:.2f} (Method: {self.raw_limit_method})")
+        logger.debug(f"Raw CPU limit determined: {self.raw_limit_value:.2f} (Method: {self.raw_limit_method})")
 
-        # --- 6. Apply Weighting (if applicable) ---
+        # --- 7. Apply Weighting (if applicable) ---
         final_effective_cores = raw_limit
         final_method = raw_method
 
@@ -394,7 +540,7 @@ class SystemResourceProbe:
         self.effective_cores = final_effective_cores
         self.detection_method = final_method  # The method for the final value
 
-        logger.info(
+        logger.debug(
             f"Effective CPU core limit determined: {self.effective_cores:.2f} " f"(Method: {self.detection_method})"
         )
 
@@ -402,6 +548,18 @@ class SystemResourceProbe:
         """Returns the primary result: the effective core limit, potentially weighted."""
         return self.effective_cores
 
+    @property
+    def total_memory_mb(self) -> Optional[float]:
+        """Returns the effective memory limit in megabytes."""
+        if self.effective_memory_bytes is not None:
+            return self.effective_memory_bytes / (1024 * 1024)
+        return None
+
+    @property
+    def cpu_count(self) -> Optional[float]:
+        """Returns the effective CPU count for compatibility."""
+        return self.effective_cores
+
     def get_details(self) -> Dict[str, Any]:
         """Returns a dictionary with all detected information."""
         # Calculate full system weighted potential for info
@@ -426,5 +584,12 @@ class SystemResourceProbe:
             "cgroup_shares": self.cgroup_shares,
             "cgroup_usage_total_us": self.cgroup_usage_total_us,
             "cgroup_usage_percpu_us": self.cgroup_usage_percpu_us,
+            # Memory information
+            "effective_memory_bytes": self.effective_memory_bytes,
+            "effective_memory_mb": self.total_memory_mb,
+            "memory_detection_method": self.memory_detection_method,
+            "os_total_memory_bytes": self.os_total_memory_bytes,
+            "cgroup_memory_limit_bytes": self.cgroup_memory_limit_bytes,
+            "cgroup_memory_usage_bytes": self.cgroup_memory_usage_bytes,
             "platform": platform.system(),
         }

{nv_ingest_api-2025.8.14.dev20250814.dist-info → nv_ingest_api-2025.8.16.dev20250816.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nv-ingest-api
-Version: 2025.8.14.dev20250814
+Version: 2025.8.16.dev20250816
 Summary: Python module with core document ingestion functions.
 Author-email: Jeremy Dyer <jdyer@nvidia.com>
 License:                                  Apache License

{nv_ingest_api-2025.8.14.dev20250814.dist-info → nv_ingest_api-2025.8.16.dev20250816.dist-info}/RECORD

@@ -7,7 +7,7 @@ nv_ingest_api/interface/transform.py,sha256=g6YnFR7TpEU0xNtzCvv6kqnFbuCwQ6vRMjjB
 nv_ingest_api/interface/utility.py,sha256=AL4l0cJNvTjG1MAe1YNTk1jbbPED3g4HCewzx6Ffcio,7296
 nv_ingest_api/internal/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nv_ingest_api/internal/enums/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest_api/internal/enums/common.py,sha256=lzDJ35VWfIwlL_Lx_q0dfHUuwEB7CXudHIQAilpjoRw,12611
+nv_ingest_api/internal/enums/common.py,sha256=qD2fEYyvPTmurJZZL7MXOmcgRiK5aCn8EZI4b4yjlB4,13342
 nv_ingest_api/internal/extract/__init__.py,sha256=uLsBITo_XfgbwpzqXUm1IYX6XlZrTfx6T1cIhdILwG8,140
 nv_ingest_api/internal/extract/audio/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/internal/extract/audio/audio_extraction.py,sha256=_jf_UC_FTqZr-xEpwG8edwBzdDjM01gGhqm9ulOsDcY,6973
@@ -21,7 +21,7 @@ nv_ingest_api/internal/extract/html/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQus
 nv_ingest_api/internal/extract/html/html_extractor.py,sha256=I9oWfj6_As4898GDDh0zsSuKxO3lBsvyYzhvUotjzJI,3282
 nv_ingest_api/internal/extract/image/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/internal/extract/image/chart_extractor.py,sha256=gk-O-9wjZBoaLVE_6Erb4gMwsSFk4UtPQ2QLpMCW4H4,13212
-nv_ingest_api/internal/extract/image/image_extractor.py,sha256=4tUWinuFMN3ukWa2tZa2_LtzRiTyUAUCBF6BDkUEvm0,8705
+nv_ingest_api/internal/extract/image/image_extractor.py,sha256=gBKjlx28hA_e-dupatu46YQgOHJ0DLpAWxREiLaZLyo,9039
 nv_ingest_api/internal/extract/image/infographic_extractor.py,sha256=i7zt_ow1gytU4hK2JCRg7T1wlbokaeuUpXX69LIQkzY,9687
 nv_ingest_api/internal/extract/image/table_extractor.py,sha256=O0m3N2Tz9W6X7TBI4o-rbBXc8dFOf9zSZq1v9qC1U4M,13780
 nv_ingest_api/internal/extract/image/image_helpers/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
@@ -40,12 +40,14 @@ nv_ingest_api/internal/extract/pptx/__init__.py,sha256=HIHfzSig66GT0Uk8qsGBm_f13
 nv_ingest_api/internal/extract/pptx/pptx_extractor.py,sha256=o-0P2dDyRFW37uQi_lKk6-eFozTcZvbq-2Y4I0EBMIY,7749
 nv_ingest_api/internal/extract/pptx/engines/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nv_ingest_api/internal/extract/pptx/engines/pptx_helper.py,sha256=p9-77JoG-ztMBrAFfvveSx94y0OE-Q4oCJfzw2iMVgI,29754
+nv_ingest_api/internal/meta/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
+nv_ingest_api/internal/meta/udf.py,sha256=kBBsU5DD1oZZyCMLd6TGh1sHahh9FNkj80wN_L--asU,8620
 nv_ingest_api/internal/mutate/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/internal/mutate/deduplicate.py,sha256=hmvTTGevpCtlkM_wVZSoc8-Exr6rUJwqLjoEnbPcPzY,3849
 nv_ingest_api/internal/mutate/filter.py,sha256=H-hOTBVP-zLpvQr-FoGIJKxkhtj4l_sZ9V2Fgu3rTEM,5183
 nv_ingest_api/internal/primitives/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nv_ingest_api/internal/primitives/control_message_task.py,sha256=nWVB3QsP6p8BKwHThd-SNv_zwJAEA1mKCRharuju1mc,439
-nv_ingest_api/internal/primitives/ingest_control_message.py,sha256=rvipBiiUaHuRhupFCFDCG8rv0PylSJibCiJ7rDeb98A,8514
+nv_ingest_api/internal/primitives/ingest_control_message.py,sha256=8rA0UbPDSB3avReAKNxiUa_FCy7fIQpqk6tfmcYUibA,9879
 nv_ingest_api/internal/primitives/nim/__init__.py,sha256=i_i_fBR2EcRCh2Y19DF6GM3s_Q0VPgo_thPnhEIJUyg,266
 nv_ingest_api/internal/primitives/nim/default_values.py,sha256=W92XjfyeC6uuVxut6J7p00x1kpNsnXIDb97gSVytZJk,380
 nv_ingest_api/internal/primitives/nim/nim_client.py,sha256=m8_whtmfXFGSw4-Efu938NKG97nzhygyuHB8Tq-a0Ec,16570
@@ -64,7 +66,7 @@ nv_ingest_api/internal/primitives/nim/model_interface/yolox.py,sha256=zpfEZIPctW
 nv_ingest_api/internal/primitives/tracing/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nv_ingest_api/internal/primitives/tracing/latency.py,sha256=5kVTeYRbRdTlT_aI4MeS20N_S7mqCcLqZR6YHtxhXkY,2215
 nv_ingest_api/internal/primitives/tracing/logging.py,sha256=SSzIgS7afLH-e1C7VagYDmkkA6rTXmQ-bmtLjoEguhg,3851
-nv_ingest_api/internal/primitives/tracing/tagging.py,sha256=O5dD7Z7j43nrjqn0AxhxOPm5zIyMFo0akxaWU_FguAM,7866
+nv_ingest_api/internal/primitives/tracing/tagging.py,sha256=xU534rb94uKnsSu0_DzyZcCSkIpa5SWTMxX7NSA3HoE,11671
 nv_ingest_api/internal/schemas/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/internal/schemas/extract/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/internal/schemas/extract/extract_audio_schema.py,sha256=W-nEBriqiNkjpaQ5AT_8LhtVXlW8AhlcftmoeQQtKAs,3812
@@ -82,8 +84,9 @@ nv_ingest_api/internal/schemas/message_brokers/request_schema.py,sha256=LZX_wXDx
 nv_ingest_api/internal/schemas/message_brokers/response_schema.py,sha256=4b275HlzBSzpmuE2wdoeaGKPCdKki3wuWldtRIfrj8w,727
 nv_ingest_api/internal/schemas/meta/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/internal/schemas/meta/base_model_noext.py,sha256=8hXU1uuiqZ6t8EsoZ8vlC5EFf2zSZrKEX133FcfZMwI,316
-nv_ingest_api/internal/schemas/meta/ingest_job_schema.py,sha256=ceYQjRjhBSDbbZ6q-Db7Y6GHVOvWPdGAMb3TX1vMWfY,8321
+nv_ingest_api/internal/schemas/meta/ingest_job_schema.py,sha256=En7wcLPB6hvfAXx2-xZM49wbJXmRX2Ckc0i6edqn21c,10145
 nv_ingest_api/internal/schemas/meta/metadata_schema.py,sha256=VnAzkSFat_ckI19mlwQTlFrvP6EZVCwyNl9bt51b8oU,7193
+nv_ingest_api/internal/schemas/meta/udf.py,sha256=GgzqbZOlipQgMpDhbXLqbF8xrHenj_hMNqhR_P-1ynw,779
 nv_ingest_api/internal/schemas/mutate/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/internal/schemas/mutate/mutate_image_dedup_schema.py,sha256=k1JOdlPPpsipc0XhHf-9YxJ_-W0HvpVE1ZhYmr7fzj0,395
 nv_ingest_api/internal/schemas/store/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
@@ -99,7 +102,7 @@ nv_ingest_api/internal/store/embed_text_upload.py,sha256=maxb4FPsBvWgvlrjAPEBlRZ
 nv_ingest_api/internal/store/image_upload.py,sha256=GNlY4k3pfcHv3lzXxkbmGLeHFsf9PI25bkBn6Xn9h3I,9654
 nv_ingest_api/internal/transform/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/internal/transform/caption_image.py,sha256=0ILCG2F8ESqKtZiPUM-6F1BHUflFZ76Dzi2GNzkE-lU,8517
-nv_ingest_api/internal/transform/embed_text.py,sha256=z9P5AtUo_e7s5qLlyhcUISQ1HUhBD93bdmEcXls_Fwc,19768
+nv_ingest_api/internal/transform/embed_text.py,sha256=AdUXVosBHuZadaFLi9_RVB7vB_hjTziCLUeBZ1oWGZo,20003
 nv_ingest_api/internal/transform/split_text.py,sha256=LAtInGVuydH43UwjNMQWFVC1A6NdhXP_dZup2xX4qEo,7745
 nv_ingest_api/util/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/util/control_message/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -115,7 +118,7 @@ nv_ingest_api/util/detectors/__init__.py,sha256=HIHfzSig66GT0Uk8qsGBm_f13fKYcPtI
 nv_ingest_api/util/detectors/language.py,sha256=TvzcESYY0bn0U4aLY6GjB4VaCWA6XrXxAGZbVzHTMuE,965
 nv_ingest_api/util/exception_handlers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nv_ingest_api/util/exception_handlers/converters.py,sha256=fX64p5GeoHOPbGccR2JrsKQLa0Ge9Mmi2W-9mLnVVl8,2221
-nv_ingest_api/util/exception_handlers/decorators.py,sha256=ywaqekT9g2gfDH8my7JTJdcnDYbguSzWfAQ6TVxqQvE,21459
+nv_ingest_api/util/exception_handlers/decorators.py,sha256=bDuazHYDDCTMT7TlyLIzM4iS6eB7O3xlJu3yU04Iw2k,18527
 nv_ingest_api/util/exception_handlers/detectors.py,sha256=Q1O-QOzsShPpNian2lawXVAOCstIE7nSytNw516hTg8,2288
 nv_ingest_api/util/exception_handlers/pdf.py,sha256=FUC41QJKDCfiTv-1c1_8Isxwt1xMxDZw9BN8JLEJKBw,3654
 nv_ingest_api/util/exception_handlers/schemas.py,sha256=NJngVNf9sk5Uz6CFFfkNO_LBAMt2QZUcMYGxX64oYRk,2179
@@ -125,10 +128,13 @@ nv_ingest_api/util/image_processing/processing.py,sha256=LSoDDEmahr7a-qSS12McVco
 nv_ingest_api/util/image_processing/table_and_chart.py,sha256=idCIjiLkY-usI2EARchg3omWLtIYmYA-1tdUUV2lbno,16338
 nv_ingest_api/util/image_processing/transforms.py,sha256=FBcORrvjimn3c1naaVxRMm6PMJ2Dt6Uy9AZRUxdbkR0,23829
 nv_ingest_api/util/imports/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
-nv_ingest_api/util/imports/callable_signatures.py,sha256=e2bJB1pmkN4Ee-Bf-VggOSBaQ4RXofWF5eKkWXgIj2U,1855
-nv_ingest_api/util/imports/dynamic_resolvers.py,sha256=7GByV_-8z2X0tnVoabCxVioxOP3sYMros3ZllVAW-wY,4343
+nv_ingest_api/util/imports/callable_signatures.py,sha256=ipzXNZJpfu7oeTBrQz2h6zrFVIQaqb2KBpzSuIX3u-Y,4138
+nv_ingest_api/util/imports/dynamic_resolvers.py,sha256=qy7RpmBZrXJarOQl3J7jiCKnbZMNChXTL_Z-H4c9zlc,6170
+nv_ingest_api/util/introspection/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
+nv_ingest_api/util/introspection/class_inspect.py,sha256=sEYe37ICHdhXxSbD0JTCDg3J9XIhYwCc2D4ysdT4XK4,5037
+nv_ingest_api/util/introspection/function_inspect.py,sha256=_yCUUT1x3tLCsbXmUpTv_O8Qi6af4NiCpJ1qGfcksvw,2066
 nv_ingest_api/util/logging/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-nv_ingest_api/util/logging/configuration.py,sha256=XUo7yON9V8IDPfN3x8RBwpZ3Gv4zrRq8QwsByf4dGNE,981
+nv_ingest_api/util/logging/configuration.py,sha256=05KR3LOS-PCqU-Io__iiKG_Ds730eKxciklFfNeId3w,3126
 nv_ingest_api/util/message_brokers/__init__.py,sha256=wQSlVx3T14ZgQAt-EPzEczQusXVW0W8yynnUaFFGE3s,143
 nv_ingest_api/util/message_brokers/simple_message_broker/__init__.py,sha256=WaQ3CWIpIKWEivT5kL-bkmzcSQKLGFNFHdXHUJjqZFs,325
 nv_ingest_api/util/message_brokers/simple_message_broker/broker.py,sha256=PekxaxVcAa9k1wgUtozlr04SW3sAeqYJE-wdVBZf9eo,17264
@@ -151,10 +157,12 @@ nv_ingest_api/util/service_clients/redis/redis_client.py,sha256=3NLecvIvVN1v-sA7
 nv_ingest_api/util/service_clients/rest/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nv_ingest_api/util/service_clients/rest/rest_client.py,sha256=dZ-jrk7IK7oNtHoXFSNTf7psoOpLREiLN5ezpHFW0HI,21732
 nv_ingest_api/util/string_processing/__init__.py,sha256=mkwHthyS-IILcLcL1tJYeF6mpqX3pxEw5aUzDGjTSeU,1411
+nv_ingest_api/util/string_processing/configuration.py,sha256=2HS08msccuPCT0fn_jfXRo9_M6hCZ59OxKLxG_47HRY,29888
+nv_ingest_api/util/string_processing/yaml.py,sha256=6SW2O6wbXRhGbhETMbtXjYCZn53HeCNOP6a96AaxlHs,1454
 nv_ingest_api/util/system/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-nv_ingest_api/util/system/hardware_info.py,sha256=ORZeKpH9kSGU_vuPhyBwkIiMyCViKUX2CP__MCjrfbU,19463
-nv_ingest_api-2025.8.14.dev20250814.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-nv_ingest_api-2025.8.14.dev20250814.dist-info/METADATA,sha256=w_oB62lsoQzWiyf6U0daYgCvp_VbZDlvXiCPFySHNY4,13947
-nv_ingest_api-2025.8.14.dev20250814.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-nv_ingest_api-2025.8.14.dev20250814.dist-info/top_level.txt,sha256=abjYMlTJGoG5tOdfIB-IWvLyKclw6HLaRSc8MxX4X6I,14
-nv_ingest_api-2025.8.14.dev20250814.dist-info/RECORD,,
+nv_ingest_api/util/system/hardware_info.py,sha256=1UFM8XE6M3pgQcpbVsCsqDQ7Dj-zzptL-XRE-DEu9UA,27213
+nv_ingest_api-2025.8.16.dev20250816.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+nv_ingest_api-2025.8.16.dev20250816.dist-info/METADATA,sha256=i8zdn2_kHXVUBq5VBK3skgNS6494ZHb9Dxmp8vY01qo,13947
+nv_ingest_api-2025.8.16.dev20250816.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+nv_ingest_api-2025.8.16.dev20250816.dist-info/top_level.txt,sha256=abjYMlTJGoG5tOdfIB-IWvLyKclw6HLaRSc8MxX4X6I,14
+nv_ingest_api-2025.8.16.dev20250816.dist-info/RECORD,,