remem-ai 0.6.49

use super::{extract_field, ParsedObservation, OBSERVATION_TYPES};

const INVALID_TYPE_PREVIEW_BYTES: usize = 120;

#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub(crate) enum InvalidObservationTypeDrop {
    Missing,
    Unknown,
}

#[derive(Debug, Clone, PartialEq)]
pub(crate) struct ParseObservationsOutcome {
    pub(crate) observations: Vec<ParsedObservation>,
    pub(crate) invalid_type_drops: Vec<InvalidObservationTypeDrop>,
}

impl ParseObservationsOutcome {
    pub(crate) fn had_invalid_type(&self) -> bool {
        !self.invalid_type_drops.is_empty()
    }
}

/// Find `needle` in `haystack` using ASCII case-insensitive comparison.
/// Returns the byte offset of the first match, or `None`.
pub(crate) fn find_ascii_ci(haystack: &str, needle: &str) -> Option<usize> {
    let needle = needle.as_bytes();
    haystack
        .as_bytes()
        .windows(needle.len())
        .position(|w| w.eq_ignore_ascii_case(needle))
}

fn extract_array(content: &str, array_name: &str, element_name: &str) -> Vec<String> {
    let open = format!("<{}>", array_name);
    let close = format!("</{}>", array_name);
    let Some(start) = content.find(&open) else {
        return vec![];
    };
    let start = start + open.len();
    let Some(end_rel) = content[start..].find(&close) else {
        return vec![];
    };
    let end = start + end_rel;
    let inner = &content[start..end];

    let elem_open = format!("<{}>", element_name);
    let elem_close = format!("</{}>", element_name);
    let mut results = Vec::new();
    let mut pos = 0;
    while let Some(found) = inner[pos..].find(&elem_open) {
        let value_start = pos + found + elem_open.len();
        let Some(end_rel) = inner[value_start..].find(&elem_close) else {
            break;
        };
        let value_end = value_start + end_rel;
        let value = inner[value_start..value_end].trim().to_string();
        if !value.is_empty() {
            results.push(value);
        }
        pos = value_end + elem_close.len();
    }
    results
}

/// Parse the optional `<confidence>` field. A missing field is normal model
/// output and stays silent; a present-but-unparseable value (e.g. "very high")
/// is logged at warn level before falling back, so silent quality degradation
/// stays observable (the caller substitutes the default confidence).
fn parse_confidence(content: &str) -> Option<f64> {
    let raw = extract_field(content, "confidence")?;
    match raw.parse::<f64>() {
        Ok(value) if value.is_finite() => Some(value.clamp(0.0, 1.0)),
        _ => {
            crate::log::warn(
                "observation-parse",
                &format!("invalid <confidence> value {raw:?}; falling back to default"),
            );
            None
        }
    }
}

pub fn parse_observations(text: &str) -> Vec<ParsedObservation> {
    parse_observations_with_outcome(text).observations
}

pub(crate) fn parse_observations_with_outcome(text: &str) -> ParseObservationsOutcome {
    let mut observations = Vec::new();
    let mut invalid_type_drops = Vec::new();
    let mut pos = 0;

    while let Some(tag_start_rel) = find_ascii_ci(&text[pos..], "<observation") {
        let tag_start = pos + tag_start_rel;
        let Some(open_end_rel) = text[tag_start..].find('>') else {
            break;
        };
        let content_start = tag_start + open_end_rel + 1;
        let Some(close_rel) = find_ascii_ci(&text[content_start..], "</observation>") else {
            break;
        };
        let content_end = content_start + close_rel;
        let content = &text[content_start..content_end];

        let Some(raw_type) = extract_field(content, "type") else {
            crate::log::error(
                "observation-parse",
                "dropping observation: drop_reason=missing_type raw_type=\"\"",
            );
            invalid_type_drops.push(InvalidObservationTypeDrop::Missing);
            pos = content_end + "</observation>".len();
            continue;
        };
        let obs_type = raw_type.trim().to_ascii_lowercase();
        if !OBSERVATION_TYPES.contains(&obs_type.as_str()) {
            let raw_type_preview = crate::adapter::redaction::redact_and_truncate(
                &raw_type,
                INVALID_TYPE_PREVIEW_BYTES,
            );
            crate::log::error(
                "observation-parse",
                &format!(
                    "dropping observation: drop_reason=unknown_type raw_type_preview={raw_type_preview:?} raw_type_bytes={}",
                    raw_type.len()
                ),
            );
            invalid_type_drops.push(InvalidObservationTypeDrop::Unknown);
            pos = content_end + "</observation>".len();
            continue;
        }

        let mut concepts = extract_array(content, "concepts", "concept");
        concepts.retain(|concept| !concept.eq_ignore_ascii_case(&obs_type));

        observations.push(ParsedObservation {
            obs_type,
            title: extract_field(content, "title"),
            subtitle: extract_field(content, "subtitle"),
            facts: extract_array(content, "facts", "fact"),
            narrative: extract_field(content, "narrative"),
            concepts,
            files_read: extract_array(content, "files_read", "file"),
            files_modified: extract_array(content, "files_modified", "file"),
            confidence: parse_confidence(content),
        });

        pos = content_end + "</observation>".len();
    }

    ParseObservationsOutcome {
        observations,
        invalid_type_drops,
    }
}