time_unit fix for Geologic timescale

pyleoclim/utils/datasets.py

@@ -5,6 +5,8 @@
 import pyleoclim as pyleo
 import yaml
 import pandas as pd
+
+from .tsbase import disambiguate_time_metadata
 from ..utils import jsonutils
 
 DATA_DIR = Path(__file__).parents[1].joinpath("data").resolve()
@@ -230,7 +232,15 @@ def _localname(term):
 
 
 
-def load_ics_chart_to_df(ttl_path_or_url="https://raw.githubusercontent.com/i-c-stratigraphy/chart/refs/heads/main/chart.ttl", time_units='Ma') -> pd.DataFrame:
+def load_ics_chart_to_df(ttl_path_or_url="https://raw.githubusercontent.com/i-c-stratigraphy/chart/refs/heads/main/chart.ttl",
+                         time_units='Ma', lang='en') -> pd.DataFrame:
+
+    def _first_lang(iterable, lang=None):
+        for lit in iterable:
+            if isinstance(lit, Literal) and (lang is None or lit.language == lang):
+                return lit
+        return None
+
 
     try:
         from rdflib import Graph, Namespace, URIRef, Literal
@@ -285,8 +295,10 @@ def _find_ns_for_local(graph, local):
             rank_local = _localname(rank_uri)
             Rank = rank_map.get(rank_local, rank_local)
 
-            # Name
-            name_lit = _first(g.objects(subj, SKOS.prefLabel))
+
+            name_lit = _first_lang(g.objects(subj, SKOS.prefLabel), lang)
+            if name_lit is None:
+                name_lit = _first(g.objects(subj, SKOS.prefLabel))
             Name = str(name_lit) if name_lit is not None else ""
 
             # Abbrev: skos:notation (typed literal is fine)
@@ -354,19 +366,22 @@ def boundary_ma(node_pred):
     df["Rank"] = pd.Categorical(df["Rank"], categories=type_order + sorted(set(df["Rank"]) - set(type_order)), ordered=True)
     df = df.sort_values(["Rank", "UpperBoundary"], na_position="last").reset_index(drop=True)
 
-    if time_units not in ['Ma', 'Mya', 'My']:
-        if time_units in ['a', 'ya', 'y', 'yr']:
+    time_name, time_unit = disambiguate_time_metadata(time_units)
+    print(f"Time name: {time_name}")
+    print(f"Time unit: {time_unit}")
+    if time_unit not in ['Ma']:#['Ma', 'Mya', 'My']:
+        if  time_unit in ['a', 'ya', 'y', 'yr', 'yrs', 'year', 'years']:
             df['UpperBoundary'] = df['UpperBoundary'] * 1e6
             df['LowerBoundary'] = df['LowerBoundary'] * 1e6
-        elif time_units in ['ka', 'kya', 'ky', 'kyr']:
+        elif time_unit in ['ka']:#['ka', 'kya', 'ky', 'kyr']:
             df['UpperBoundary'] = df['UpperBoundary'] * 1e3
             df['LowerBoundary'] = df['LowerBoundary'] * 1e3
             print('converted to ka')
-        elif time_units in ['Ga', 'Gya', 'Gy', 'Gyr']:
+        elif time_unit in ['Ga']:#['Ga', 'Gya', 'Gy', 'Gyr']:
             df['UpperBoundary'] = df['UpperBoundary'] * 1e-3
             df['LowerBoundary'] = df['LowerBoundary'] * 1e-3
         else:
-            raise ValueError(f"Unsupported time_units '{time_units}'. Supported: 'Ma', 'ka', 'Ga'.")
+            raise ValueError(f"Unsupported time_units '{time_unit}'. Supported: 'Ma', 'ka', 'Ga'.")
 
     return df
 

pyleoclim/utils/plotting.py

@@ -1518,9 +1518,9 @@ def text_loc(fig, ax, rect, label_text, width, yloc):
     text.remove()
     return loc
 
-def add_GTS(fig, ax, GTS_df=None, ranks=None, time_units='Ma',location='above', label_pref='full',
+def add_GTS(fig, ax, time_units, GTS_df=None, ranks=None, location='above', label_pref='full',
             allow_abbreviations=True, ax_name='gts', v_offset=0, height=.05, text_color=None, fontsize=None, edgecolor='k',
-            edgewidth=0, alpha=1,zorder=10,reverse_rank_order=True,
+            edgewidth=0, alpha=1,zorder=10,reverse_rank_order=True, lang='en',
             gts_url = "https://raw.githubusercontent.com/i-c-stratigraphy/chart/refs/heads/main/chart.ttl"):
 
     '''
@@ -1531,13 +1531,13 @@ def add_GTS(fig, ax, GTS_df=None, ranks=None, time_units='Ma',location='above',
         The figure object where the GTS will be added.
     ax : dict
         The axis where the GTS will be plotted.
+    time_units : str
+        Time units of the GTS data. Supported: 'Ma' (default), 'ka', 'Ga'.
     GTS_df : pd.DataFrame, optional
         DataFrame containing the GTS data with columns for 'Rank', 'Name', 'Abbrev', 'Color', 'UpperBoundary', 'LowerBoundary'.
         If None, the function will load the ICS chart from the provided URL.
     ranks : list of str, optional
         List of ranks to include (e.g., ['Period', 'Epoch', 'Stage']). If None, defaults to ['Period', 'Epoch', 'Stage']. Ranks are ordered from outer-most to inner-most.
-    time_units : str, optional
-        Time units of the GTS data. Supported: 'Ma' (default), 'ka', 'Ga'.
     location : str, optional
         Specifies whether to place the GTS above or below the plot. Default is 'above'.
     label_pref : str, optional
@@ -1608,7 +1608,7 @@ def add_GTS(fig, ax, GTS_df=None, ranks=None, time_units='Ma',location='above',
         assert all(duration >= 0), "GTS_df 'LowerBoundary' must be greater than (further back in time) or equal to 'UpperBoundary (more modern)'"
 
     if GTS_df is None:
-        GTS_df = datasets.load_ics_chart_to_df(gts_url, time_units=time_units)
+        GTS_df = datasets.load_ics_chart_to_df(gts_url, time_units=time_units, lang=lang)
 
     if ranks is None:
         ranks = ['Period', 'Epoch', 'Stage']