Implement final loni generation logic

oonidata/analysis/website_experiment_results.py

@@ -9,13 +9,6 @@
 log = logging.getLogger("oonidata.analysis")
 
 
-class LoniValue(NamedTuple):
-    ok_value: float
-    down: Dict[str, float]
-    blocked: Dict[str, float]
-    blocking_scope: Optional[str]
-
-
 def map_analysis_to_target_name(analysis):
     # Poormans mapping to target name
     # TODO(arturo) we eventually want to look these up in some sort of database that groups together related domains
@@ -60,10 +53,21 @@ def sum(self) -> float:
         return s
 
 
+@dataclass
+class LoNI:
+    ok_final: float
+
+    ok: OutcomeSpace
+    down: OutcomeSpace
+    blocked: OutcomeSpace
+    blocking_scope: Optional[str]
+
+
 def calculate_web_loni(
     web_analysis: WebAnalysis,
-) -> Tuple[LoniValue, List[str]]:
+) -> Tuple[LoNI, List[str]]:
     ok_value = 0
+    ok = OutcomeSpace()
     down = OutcomeSpace()
     blocked = OutcomeSpace()
 
@@ -332,6 +336,7 @@ def calculate_web_loni(
     # final analysis to blocked and down dictionaries.
     blocked.dns = OutcomeStatus(key=blocked_key, value=blocked_value)
     down.dns = OutcomeStatus(key=down_key, value=down_value)
+    ok.dns = OutcomeStatus(key="dns", value=ok_value)
     assert (
         blocked.sum() + down.sum() + ok_value == 1
     ), f"{blocked} + {down} + {ok_value} != 1"
@@ -343,10 +348,11 @@ def calculate_web_loni(
         # TODO(arturo): do we want to have different thresholds here?
         analysis_transcript.append("ok_value < 0.5")
         return (
-            LoniValue(
-                ok_value=ok_value,
-                blocked=blocked.to_dict(),
-                down=down.to_dict(),
+            LoNI(
+                ok_final=ok_value,
+                ok=ok,
+                blocked=blocked,
+                down=down,
                 blocking_scope=blocking_scope,
             ),
             analysis_transcript,
@@ -421,6 +427,8 @@ def calculate_web_loni(
 
         blocked.tcp = OutcomeStatus(key=blocked_key, value=blocked_value * ok_value)
         down.tcp = OutcomeStatus(key=down_key, value=down_value * ok_value)
+        # TODO(arturo): double check this is correct
+        ok.tcp = OutcomeStatus(key="tcp", value=1 - (blocked.sum() + down.sum()))
 
     old_ok_value = ok_value
     ok_value = 1 - (blocked.sum() + down.sum())
@@ -437,10 +445,11 @@ def calculate_web_loni(
             f"ok_value < 0.5 # OK went from {old_ok_value} -> {ok_value}"
         )
         return (
-            LoniValue(
-                ok_value=ok_value,
-                blocked=blocked.to_dict(),
-                down=down.to_dict(),
+            LoNI(
+                ok_final=ok_value,
+                ok=ok,
+                blocked=blocked,
+                down=down,
                 blocking_scope=blocking_scope,
             ),
             analysis_transcript,
@@ -528,6 +537,8 @@ def calculate_web_loni(
 
         blocked.tls = OutcomeStatus(key=blocked_key, value=blocked_value * ok_value)
         down.tls = OutcomeStatus(key=down_key, value=down_value * ok_value)
+        # TODO(arturo): double check this is correct
+        ok.tls = OutcomeStatus(key="tls", value=1 - (blocked.sum() + down.sum()))
 
     old_ok_value = ok_value
     ok_value = 1 - (blocked.sum() + down.sum())
@@ -542,10 +553,11 @@ def calculate_web_loni(
             f"ok_value < 0.5 # OK went from {old_ok_value} -> {ok_value}"
         )
         return (
-            LoniValue(
-                ok_value=ok_value,
-                blocked=blocked.to_dict(),
-                down=down.to_dict(),
+            LoNI(
+                ok_final=ok_value,
+                ok=ok,
+                blocked=blocked,
+                down=down,
                 blocking_scope=blocking_scope,
             ),
             analysis_transcript,
@@ -707,11 +719,15 @@ def calculate_web_loni(
                 )
             blocked.tls = OutcomeStatus(key=blocked_key, value=blocked_value * ok_value)
             down.tls = OutcomeStatus(key=down_key, value=down_value * ok_value)
+            # TODO(arturo): double check this is correct
+            ok.tls = OutcomeStatus(key="tls", value=1 - (blocked.sum() + down.sum()))
         else:
             blocked.http = OutcomeStatus(
                 key=blocked_key, value=blocked_value * ok_value
             )
             down.http = OutcomeStatus(key=down_key, value=down_value * ok_value)
+            # TODO(arturo): double check this is correct
+            ok.http = OutcomeStatus(key="http", value=1 - (blocked.sum() + down.sum()))
 
     old_ok_value = ok_value
     ok_value = 1 - (blocked.sum() + down.sum())
@@ -720,10 +736,11 @@ def calculate_web_loni(
     ), f"{blocked} + {down} + {ok_value} != 1"
 
     return (
-        LoniValue(
-            ok_value=ok_value,
-            blocked=blocked.to_dict(),
-            down=down.to_dict(),
+        LoNI(
+            ok_final=ok_value,
+            ok=ok,
+            blocked=blocked,
+            down=down,
             blocking_scope=blocking_scope,
         ),
         analysis_transcript,
@@ -750,21 +767,145 @@ def make_website_experiment_results(
     target_domain_name = first_analysis.target_domain_name
     target_detail = first_analysis.target_detail
 
-    loni_ok_value = 0
+    analysis_transcript_list = []
+    loni_list = []
+    loni_blocked_list = []
+    loni_down_list = []
+    loni_ok_list = []
+    for wa in web_analysis:
+        loni, analysis_transcript = calculate_web_loni(wa)
+        analysis_transcript_list.append(analysis_transcript)
+        loni_list.append(loni)
+        loni_blocked_list.append(loni.blocked)
+        loni_down_list.append(loni.down)
+        loni_ok_list.append(loni.ok)
+
+    final_blocked = OutcomeSpace()
+    final_down = OutcomeSpace()
+    final_ok = OutcomeSpace()
+    ok_value = 0
+    blocking_scope = None
+
+    # TODO(arturo): this section needs to be formalized and verified a bit more
+    # in depth. Currently it's just a prototype to start seeing how the data
+    # looks like.
 
-    loni_down_keys = []
-    loni_down_values = []
+    ### FINAL DNS
+    max_dns_blocked = max(
+        map(lambda x: x.dns, loni_blocked_list), key=lambda d: d.value if d else 0
+    )
+    max_dns_down = max(
+        map(lambda x: x.dns, loni_down_list), key=lambda d: d.value if d else 0
+    )
+    min_dns_ok = min(
+        map(lambda x: x.dns, loni_ok_list), key=lambda d: d.value if d else 0
+    )
+    ok_value = min_dns_ok
+    final_ok.dns = OutcomeStatus(key="dns", value=min_dns_ok)
+    final_blocked.dns = OutcomeStatus(key=max_dns_blocked.key, value=max_dns_blocked)
+    final_down.dns = OutcomeStatus(
+        # TODO(arturo): this is overestimating blocking.
+        key=max_dns_down.key,
+        value=1 - (min_dns_ok + max_dns_blocked),
+    )
+    if max_dns_blocked.blocking_scope:
+        blocking_scope = max_dns_blocked.blocking_scope
 
-    loni_blocked_keys = []
-    loni_blocked_values = []
+    ### FINAL TCP
+    max_tcp_blocked = max(
+        map(lambda x: x.tcp, loni_blocked_list), key=lambda d: d.value if d else 0
+    )
+    max_tcp_down = max(
+        map(lambda x: x.tcp, loni_down_list), key=lambda d: d.value if d else 0
+    )
+    min_tcp_ok = min(
+        map(lambda x: x.tcp, loni_ok_list), key=lambda d: d.value if d else 0
+    )
+    final_blocked.tcp = OutcomeStatus(
+        key=max_tcp_blocked.key, value=max_tcp_blocked.value * ok_value
+    )
+    final_down.tcp = OutcomeStatus(
+        key=max_tcp_down.value,
+        value=1 - (min_tcp_ok.value + max_tcp_blocked.value) * ok_value,
+    )
+    final_ok.tcp = OutcomeStatus(key="tcp", value=min_tcp_ok.value)
 
-    analysis_transcript = []
+    old_ok_value = ok_value
+    ok_value = 1 - (final_blocked.sum() + final_down.sum())
+    log.info(f"TCP done {old_ok_value} -> {ok_value}")
 
-    is_anomaly = None
-    is_confirmed = None
+    ### FINAL TLS
+    max_tls_blocked = max(
+        map(lambda x: x.tls, loni_blocked_list), key=lambda d: d.value if d else 0
+    )
+    max_tls_down = max(
+        map(lambda x: x.tls, loni_down_list), key=lambda d: d.value if d else 0
+    )
+    min_tls_ok = min(
+        map(lambda x: x.tls, loni_ok_list), key=lambda d: d.value if d else 0
+    )
+    final_blocked.tls = OutcomeStatus(
+        key=max_tls_blocked.key, value=max_tls_blocked.value * ok_value
+    )
+    final_down.tls = OutcomeStatus(
+        key=max_tls_down.value,
+        value=1 - (min_tls_ok.value + max_tls_blocked.value) * ok_value,
+    )
+    final_ok.tls = OutcomeStatus(key="http", value=min_tls_ok.value)
+    old_ok_value = ok_value
+    ok_value = 1 - (final_blocked.sum() + final_down.sum())
+    log.info(f"TLS done {old_ok_value} -> {ok_value}")
 
-    for wa in web_analysis:
-        loni = calculate_web_loni(wa)
+    ### FINAL HTTP
+    max_http_blocked = max(
+        map(lambda x: x.http, loni_blocked_list), key=lambda d: d.value if d else 0
+    )
+    max_http_down = max(
+        map(lambda x: x.http, loni_down_list), key=lambda d: d.value if d else 0
+    )
+    min_http_ok = min(
+        map(lambda x: x.http, loni_ok_list), key=lambda d: d.value if d else 0
+    )
+    final_blocked.http = OutcomeStatus(
+        key=max_http_blocked.key, value=max_http_blocked.value * ok_value
+    )
+    final_down.http = OutcomeStatus(
+        key=max_http_down.value,
+        value=1 - (min_http_ok.value + max_http_blocked.value) * ok_value,
+    )
+    final_ok.http = OutcomeStatus(key="http", value=min_http_ok.value)
+    if max_http_blocked.blocking_scope:
+        if blocking_scope is not None:
+            log.warning(f"overwriting blocking_scope key: {blocking_scope}")
+        blocking_scope = max_http_blocked.blocking_scope
+
+    old_ok_value = ok_value
+    ok_value = 1 - (final_blocked.sum() + final_down.sum())
+    log.info(f"HTTP done {old_ok_value} -> {ok_value}")
+
+    final_loni = LoNI(
+        ok_final=ok_value,
+        ok=final_ok,
+        down=final_down,
+        blocked=final_blocked,
+        blocking_scope=blocking_scope,
+    )
+
+    loni_ok_value = final_loni.ok_final
+
+    loni_down = final_loni.down.to_dict()
+    loni_down_keys, loni_down_values = list(loni_down.keys()), list(loni_down.values())
+
+    loni_blocked = final_loni.down.to_dict()
+    loni_blocked_keys, loni_blocked_values = list(loni_blocked.keys()), list(
+        loni_blocked.values()
+    )
+
+    loni_ok = final_loni.ok.to_dict()
+    loni_ok_keys, loni_ok_values = list(loni_ok.keys()), list(loni_ok.values())
+
+    is_anomaly = loni_ok_value < 0.6
+    is_confirmed = final_loni.blocked.sum() > 0.9
 
     er = ExperimentResultInstant(
         experiment_result_id=experiment_result_id,
@@ -792,7 +933,9 @@ def make_website_experiment_results(
         loni_down_values=loni_down_values,
         loni_blocked_keys=loni_blocked_keys,
         loni_blocked_values=loni_blocked_values,
-        analysis_transcript=analysis_transcript,
+        loni_ok_keys=loni_ok_keys,
+        loni_ok_values=loni_ok_values,
+        analysis_transcript_list=analysis_transcript_list,
         measurement_count=1,
         observation_count=len(web_analysis),
         vp_count=1,

oonidata/models/experiment_result.py

@@ -163,9 +163,12 @@ class ExperimentResultInstant:
     loni_blocked_keys: List[str]
     loni_blocked_values: List[float]
 
+    loni_ok_keys: List[str]
+    loni_ok_values: List[float]
+
     # Inside this string we include a representation of the logic that lead us
     # to produce the above loni values
-    analysis_transcript: Optional[List[str]]
+    analysis_transcript_list: Optional[List[List[str]]]
 
     # Number of measurements used to produce this experiment result
     measurement_count: int