chartlet 0.18.0

use std::collections::BTreeSet;

use serde::{Deserialize, Serialize};

use super::{
    ChartSpec, ChartType, MAX_DATA_POINTS, MAX_SERIES, Stack, validate_number, validate_text,
};
use crate::error::{ChartError, ChartWarning};

#[derive(Debug, Clone, Serialize, Deserialize)]
#[serde(deny_unknown_fields)]
pub struct DataPoint {
    pub label: String,
    pub value: Option<f64>,
    /// The group a bar belongs to: one palette color and one legend entry per group. Either every
    /// point names a group or none does. Bar charts only.
    #[serde(default, skip_serializing_if = "Option::is_none")]
    pub group: Option<String>,
    /// The lower end of an error bar through the end of the bar, such as a confidence interval.
    /// Bar charts of one series only; `upper` goes with it, and the value lies between them.
    #[serde(default, skip_serializing_if = "Option::is_none")]
    pub lower: Option<f64>,
    #[serde(default, skip_serializing_if = "Option::is_none")]
    pub upper: Option<f64>,
}

/// One named series with one value per category; `None` marks a missing value.
#[derive(Debug, Clone, Serialize, Deserialize)]
#[serde(deny_unknown_fields)]
pub struct SeriesSpec {
    pub name: String,
    pub values: Vec<Option<f64>>,
    /// The lower and upper end of an error bar through each value, such as a confidence
    /// interval: both or neither, one entry per category, `null` where the value is missing.
    #[serde(default, skip_serializing_if = "Vec::is_empty")]
    pub lower: Vec<Option<f64>>,
    #[serde(default, skip_serializing_if = "Vec::is_empty")]
    pub upper: Vec<Option<f64>>,
}

impl ChartSpec {
    /// The groups of a single-series bar chart in the order they first appear; empty without
    /// groups.
    pub(crate) fn data_groups(&self) -> Vec<&str> {
        let mut groups: Vec<&str> = Vec::new();
        for group in self.data.iter().filter_map(|point| point.group.as_deref()) {
            if !groups.contains(&group) {
                groups.push(group);
            }
        }
        groups
    }

    /// A bar chart whose points name groups, as the chart it is drawn as: one series per group,
    /// each with a value only where the group has a bar, stacked so that every category keeps one
    /// full-width bar and every group gets its color and legend entry. The description and the
    /// table keep the chart as it was written.
    pub(crate) fn group_view(&self) -> Option<Self> {
        let groups = self.data_groups();
        // One group is one color: the bars are drawn as they are.
        if groups.len() < 2 {
            return None;
        }
        let mut view = self.clone();
        view.categories = self.data.iter().map(|point| point.label.clone()).collect();
        view.series = groups
            .iter()
            .map(|group| SeriesSpec {
                lower: Vec::new(),
                upper: Vec::new(),
                name: (*group).to_owned(),
                values: self
                    .data
                    .iter()
                    .map(|point| {
                        if point.group.as_deref() == Some(*group) {
                            point.value
                        } else {
                            None
                        }
                    })
                    .collect(),
            })
            .collect();
        view.data.clear();
        view.stack = Some(Stack::Normal);
        Some(view)
    }

    /// Whether the bars or points carry error bars.
    pub(crate) fn has_error_bars(&self) -> bool {
        self.data.iter().any(|point| point.lower.is_some())
            || self.series.iter().any(|series| !series.lower.is_empty())
    }

    /// The error bar of the value of series `series_index` in category `index`: the data point of
    /// a single series, or the series' own bounds.
    pub(crate) fn error_bounds(&self, index: usize, series_index: usize) -> Option<(f64, f64)> {
        if self.series.is_empty() {
            return self
                .data
                .get(index)
                .and_then(|point| point.lower.zip(point.upper));
        }
        let series = self.series.get(series_index)?;
        (*series.lower.get(index)?).zip(*series.upper.get(index)?)
    }

    /// Every error bar limit of the chart, for the value scale.
    pub(crate) fn error_limits(&self) -> impl Iterator<Item = f64> + '_ {
        let single = self
            .data
            .iter()
            .flat_map(|point| [point.lower, point.upper].into_iter().flatten());
        let several = self
            .series
            .iter()
            .flat_map(|series| series.lower.iter().chain(&series.upper).flatten().copied());
        single.chain(several)
    }

    /// Error bars of the series of a chart with several: `lower` and `upper` together, one entry
    /// for every category, around every value, on a bar or a line chart without a stack.
    fn validate_series_bounds(&self) -> Result<(), ChartError> {
        for (series_index, series) in self.series.iter().enumerate() {
            if series.lower.is_empty() && series.upper.is_empty() {
                continue;
            }
            let path = format!("/series/{series_index}");
            if !matches!(self.chart_type, ChartType::Bar | ChartType::Line) || self.stack.is_some()
            {
                return Err(ChartError::new(
                    "option_not_supported",
                    format!("{path}/lower"),
                    "error bars belong to a bar chart or a line chart without a stack",
                ));
            }
            for (name, bounds) in [("lower", &series.lower), ("upper", &series.upper)] {
                if bounds.len() != self.categories.len() {
                    return Err(ChartError::new(
                        if bounds.is_empty() {
                            "missing_bounds"
                        } else {
                            "series_length_mismatch"
                        },
                        format!("{path}/{name}"),
                        format!(
                            "give lower and upper together, {} entries each, one per category",
                            self.categories.len()
                        ),
                    ));
                }
            }
            for (index, value) in series.values.iter().enumerate() {
                let bounds = (series.lower[index], series.upper[index]);
                let (Some(value), (Some(lower), Some(upper))) = (value, bounds) else {
                    if value.is_some() {
                        return Err(ChartError::new(
                            "missing_bounds",
                            format!("{path}/lower/{index}"),
                            "give every value both lower and upper, or null where the value is null",
                        ));
                    }
                    continue;
                };
                validate_number(lower, &format!("{path}/lower/{index}"))?;
                validate_number(upper, &format!("{path}/upper/{index}"))?;
                if !(lower..=upper).contains(value) {
                    return Err(ChartError::new(
                        "invalid_bounds",
                        format!("{path}/lower/{index}"),
                        "the value must lie between lower and upper",
                    ));
                }
            }
        }
        Ok(())
    }

    /// Error bars: `lower` and `upper` together on every point or on none, around the value, on
    /// a bar chart without groups.
    fn validate_error_bars(&self) -> Result<(), ChartError> {
        self.validate_series_bounds()?;
        let used = self
            .data
            .iter()
            .any(|point| point.lower.is_some() || point.upper.is_some());
        if !used {
            return Ok(());
        }
        for (index, point) in self.data.iter().enumerate() {
            let path = format!("/data/{index}");
            if !matches!(self.chart_type, ChartType::Bar | ChartType::Line) {
                return Err(ChartError::new(
                    "option_not_supported",
                    format!("{path}/lower"),
                    "error bars belong to a bar chart or a line chart",
                ));
            }
            if point.group.is_some() {
                return Err(ChartError::new(
                    "option_not_supported",
                    format!("{path}/group"),
                    "a bar chart takes error bars or groups, not both",
                ));
            }
            let (Some(lower), Some(upper)) = (point.lower, point.upper) else {
                return Err(ChartError::new(
                    "missing_bounds",
                    path,
                    "give every point both lower and upper, or none",
                ));
            };
            validate_number(lower, &format!("{path}/lower"))?;
            validate_number(upper, &format!("{path}/upper"))?;
            let value = point.value.expect("bar points have values");
            if !(lower..=upper).contains(&value) {
                return Err(ChartError::new(
                    "invalid_bounds",
                    format!("{path}/lower"),
                    "the value must lie between lower and upper",
                ));
            }
        }
        Ok(())
    }

    /// Either every point names a group or none does, and there are no more groups than palette
    /// colors; only bar charts have groups.
    fn validate_data_groups(&self) -> Result<(), ChartError> {
        let grouped = self.data.iter().any(|point| point.group.is_some());
        let mut groups: Vec<&str> = Vec::new();
        for (index, point) in self.data.iter().enumerate() {
            let path = format!("/data/{index}/group");
            let Some(group) = point.group.as_deref() else {
                if grouped {
                    return Err(ChartError::new(
                        "missing_group",
                        path,
                        "give every point a group, or none",
                    ));
                }
                continue;
            };
            if self.chart_type != ChartType::Bar {
                return Err(ChartError::new(
                    "option_not_supported",
                    path,
                    "groups color the bars of a bar chart",
                ));
            }
            validate_text(group, &path, 60)?;
            if !groups.contains(&group) {
                groups.push(group);
            }
            if groups.len() > MAX_SERIES {
                return Err(ChartError::new(
                    "too_many_series",
                    path,
                    format!("at most {MAX_SERIES} groups are supported, one per palette color"),
                ));
            }
        }
        Ok(())
    }

    pub(super) fn validate_data(&self) -> Result<Vec<ChartWarning>, ChartError> {
        self.validate_stack()?;
        if self.categories.is_empty() && self.series.is_empty() {
            self.validate_points()
        } else {
            self.validate_series()
        }
    }

    fn validate_series_shape(&self) -> Result<(), ChartError> {
        if !self.data.is_empty() {
            return Err(ChartError::new(
                "conflicting_data_shape",
                "/data",
                "use either data or categories with series, not both",
            ));
        }
        if self.categories.is_empty() {
            return Err(ChartError::new(
                "empty_data",
                "/categories",
                "provide at least one category",
            ));
        }
        if self.categories.len() > MAX_DATA_POINTS {
            return Err(ChartError::new(
                "too_many_data_points",
                "/categories",
                format!("at most {MAX_DATA_POINTS} categories are supported"),
            ));
        }
        if self.series.is_empty() {
            return Err(ChartError::new(
                "empty_data",
                "/series",
                "provide at least one series",
            ));
        }
        if self.series.len() > MAX_SERIES {
            return Err(ChartError::new(
                "too_many_series",
                "/series",
                format!("at most {MAX_SERIES} series are supported"),
            ));
        }
        Ok(())
    }

    /// A stack needs several series; a percent stack takes no negative values, since a share of
    /// a total cannot be negative.
    fn validate_stack(&self) -> Result<(), ChartError> {
        if self.stack.is_none() {
            return Ok(());
        }
        if self.series.len() < 2 {
            return Err(ChartError::new(
                "option_not_supported",
                "/stack",
                "stacking needs categories with at least two series",
            ));
        }
        if self.stack == Some(Stack::Percent) {
            for (series_index, series) in self.series.iter().enumerate() {
                if let Some(value_index) = series
                    .values
                    .iter()
                    .position(|value| value.is_some_and(|value| value < 0.0))
                {
                    return Err(ChartError::new(
                        "invalid_value",
                        format!("/series/{series_index}/values/{value_index}"),
                        "a percent stack shows shares of a total and takes no negative values",
                    ));
                }
            }
        }
        Ok(())
    }

    fn validate_series(&self) -> Result<Vec<ChartWarning>, ChartError> {
        self.validate_series_shape()?;
        let mut labels = BTreeSet::new();
        for (index, category) in self.categories.iter().enumerate() {
            let path = format!("/categories/{index}");
            validate_text(category, &path, 200)?;
            if !labels.insert(category.as_str()) {
                return Err(ChartError::new(
                    "duplicate_label",
                    path,
                    "categories must be unique",
                ));
            }
        }

        let mut names = BTreeSet::new();
        for (series_index, series) in self.series.iter().enumerate() {
            let name_path = format!("/series/{series_index}/name");
            validate_text(&series.name, &name_path, 100)?;
            if !names.insert(series.name.as_str()) {
                return Err(ChartError::new(
                    "duplicate_series",
                    name_path,
                    "series names must be unique",
                ));
            }
            if series.values.len() != self.categories.len() {
                return Err(ChartError::new(
                    "series_length_mismatch",
                    format!("/series/{series_index}/values"),
                    format!(
                        "expected {} values, one per category; use null for a missing value",
                        self.categories.len()
                    ),
                ));
            }
            for (value_index, value) in series.values.iter().enumerate() {
                if let Some(value) = value {
                    validate_number(
                        *value,
                        &format!("/series/{series_index}/values/{value_index}"),
                    )?;
                }
            }
        }
        self.validate_series_bounds()?;
        if self
            .series
            .iter()
            .flat_map(|series| &series.values)
            .all(Option::is_none)
        {
            return Err(ChartError::new(
                "empty_series",
                "/series",
                "provide at least one numeric value",
            ));
        }

        let mut warnings = Vec::new();
        if self.categories.len() > 16 {
            warnings.push(ChartWarning::new(
                "dense_chart",
                "/categories",
                "more than 16 categories can be difficult to read at the configured size",
            ));
        }
        Ok(warnings)
    }

    fn validate_points(&self) -> Result<Vec<ChartWarning>, ChartError> {
        if self.data.is_empty() {
            return Err(ChartError::new(
                "empty_data",
                "/data",
                "provide at least one data point",
            ));
        }
        if self.data.len() > MAX_DATA_POINTS {
            return Err(ChartError::new(
                "too_many_data_points",
                "/data",
                format!("at most {MAX_DATA_POINTS} data points are supported"),
            ));
        }

        let mut labels = BTreeSet::new();
        let mut warnings = Vec::new();
        for (index, point) in self.data.iter().enumerate() {
            let label_path = format!("/data/{index}/label");
            validate_text(&point.label, &label_path, 200)?;
            if let Some(value) = point.value {
                validate_number(value, &format!("/data/{index}/value"))?;
            } else if self.chart_type == ChartType::Bar {
                return Err(ChartError::new(
                    "missing_bar_value",
                    format!("/data/{index}/value"),
                    "bar charts require a numeric value for every category",
                ));
            }
            if !labels.insert(point.label.as_str()) {
                return Err(ChartError::new(
                    "duplicate_label",
                    label_path,
                    "labels must be unique in a single-series chart",
                ));
            }
        }

        self.validate_data_groups()?;
        self.validate_error_bars()?;
        if self.data.len() > 16 {
            warnings.push(ChartWarning::new(
                "dense_chart",
                "/data",
                "more than 16 categories can be difficult to read at the configured size",
            ));
        }
        Ok(warnings)
    }
}