Fail on empty line in species list + trim (#790)

Author: Josef-Haupt

birdnet_analyzer/analyze/core.py

@@ -257,7 +257,7 @@ def _set_params(
                 if os.path.isdir(cfg.SPECIES_LIST_FILE):
                     cfg.SPECIES_LIST_FILE = os.path.join(cfg.SPECIES_LIST_FILE, "species_list.txt")
 
-            cfg.SPECIES_LIST = read_lines(cfg.SPECIES_LIST_FILE)
+            cfg.SPECIES_LIST = read_lines(cfg.SPECIES_LIST_FILE, trim=True, fail_on_blank_lines=True)
         else:
             cfg.SPECIES_LIST_FILE = None
             cfg.SPECIES_LIST = get_species_list(cfg.LATITUDE, cfg.LONGITUDE, cfg.WEEK, cfg.LOCATION_FILTER_THRESHOLD)
@@ -274,17 +274,17 @@ def _set_params(
                 cfg.LABELS_FILE = None
                 cfg.LABELS = None
             else:
-                cfg.LABELS = read_lines(cfg.LABELS_FILE)
+                cfg.LABELS = read_lines(cfg.LABELS_FILE, fail_on_blank_lines=True)
         else:
             cfg.APPLY_SIGMOID = False
             # our output format
             cfg.LABELS_FILE = os.path.join(custom_classifier, "labels", "label_names.csv")
 
             if not os.path.isfile(cfg.LABELS_FILE):
                 cfg.LABELS_FILE = os.path.join(custom_classifier, "assets", "label.csv")
-                cfg.LABELS = read_lines(cfg.LABELS_FILE)
+                cfg.LABELS = read_lines(cfg.LABELS_FILE, fail_on_blank_lines=True)
             else:
-                cfg.LABELS = [line.split(",")[1] for line in read_lines(cfg.LABELS_FILE)]
+                cfg.LABELS = [line.split(",")[1] for line in read_lines(cfg.LABELS_FILE, fail_on_blank_lines=True)]
 
     if cfg.LABELS_FILE:
         lfile = os.path.join(cfg.TRANSLATED_LABELS_PATH, os.path.basename(cfg.LABELS_FILE).replace(".txt", f"_{locale}.txt"))

birdnet_analyzer/utils.py

@@ -144,7 +144,7 @@ def collect_all_files(path: str, filetypes: list[str], pattern: str = ""):
     return sorted(files)
 
 
-def read_lines(path: str | Path | None) -> list[str]:
+def read_lines(path: str | Path | None, trim: bool = False, fail_on_blank_lines: bool = False) -> list[str]:
     """Reads the lines into a list.
 
     Opens the file and reads its contents into a list.
@@ -156,7 +156,20 @@ def read_lines(path: str | Path | None) -> list[str]:
     Returns:
         A list of all species inside the file.
     """
-    return Path(path).read_text(encoding="utf-8").splitlines() if path else []
+
+    if not path:
+        return []
+
+    lines = Path(path).read_text(encoding="utf-8").splitlines()
+    cleaned_lines = []
+
+    for line in lines:
+        if not line and fail_on_blank_lines:
+            raise ValueError(f"Blank lines are not allowed in species list\nFile: {path}")
+
+        cleaned_lines.append(line.strip() if trim else line)
+
+    return cleaned_lines
 
 
 def list_subdirectories(path: str):